A 31-operation map of how minds actually think — the irreducible mental operations grouped into six families and one generative loop (DISCERN → MODEL → INQUIRE → CREATE → ENACT, riding on AGENCY) — and the blueprint for a school that installs primitives instead of transmitting facts, in an age when intelligence itself has become a generator you can buy by the token.

The mind is not a database. We have built every school, every exam, and most of our private sense of being “smart” on the opposite assumption — that intelligence is a quantity of stored content, that the educated person is the one who has accumulated the most facts, definitions, dates, and procedures, and that learning is the slow transfer of that content from a book or a teacher into a head. This is the storage theory of mind, and it is wrong in the way that a wrong map is worse than no map. It is wrong not because facts are useless but because facts are not what thinking is made of. Thinking is made of operations.

A more honest description: the mind is a generator. It takes a base of knowledge as raw material, a context as its frame, a purpose as its function, and a procedure as its algorithm — and it produces an output: an understanding, a question, an idea, a decision, an act. Knowledge is the fuel. The generator is the asset. When you admire a brilliant person, you are not admiring the size of their warehouse; you are admiring the quality and the speed of their generators — the way they locate exactly what they do not know, find the mechanism under a surface, recombine two distant concepts into something new, and convert the result into a decision before lunch. What we call intelligence is a stack of generators running well.

This article names those generators, and gives them a more precise name still: Cognitive Primitives. A primitive, in the sense borrowed from computer science, is an irreducible operation you compose other things out of. All of software — every operating system, every database, every model — is built by composing a small set of primitives: read, write, branch, loop, map, fold. The expressive infinity of code does not come from an infinite vocabulary; it comes from the composition of a finite, small, well-chosen set of operations. The claim of this article is that the mind works the same way. You do not have a thousand separate “skills.” You have a few dozen primitives, and everything you have ever called talent, insight, wisdom, or genius is those primitives composed — chained, nested, and run in the right order.

There are 31 of them. They fall into six families, and the six families are not an arbitrary filing system: they are the phases of a single recurring loop that the mind runs whether it is learning or creating, whether a child is grasping fractions or a founder is designing a company. The loop is:

DISCERN → MODEL → INQUIRE → CREATE → ENACT — all riding on an AGENCY substrate, the self that powers, fuels, and integrates the whole cycle.

We will call this loop the Generative Loop, and the full set the Primitive Stack. The loop is the deep structure; the 31 primitives are its parts; the composition of those parts is thought itself.

Three ideas have to be installed before the catalogue makes sense, because they are the load-bearing walls of the whole framework.

First: concepts are themselves generators. This is the hidden engine of all abstraction, and most people never notice it. Take the bare word generator. To hold that concept is to hold a small machine with four sockets — it has an output, a context, a function, and an algorithm. Once you possess that machine, you can drop almost anything into it: a bubble-blower is a generator of bubbles; a car engine is a generator of motion and exhaust; a school is a generator of citizens; a brain is a generator of thoughts. The concept did not just describe those things — it gave you a new kind of relationship you can now perceive across all of them. This is the secret of conceptual depth: the more concepts you truly hold, the more kinds of relationship you can generate, and therefore the more of the world you can think about. Encyclopedic knowledge adds rows to a table. A genuinely held concept adds a new column — a new axis along which all rows can suddenly be compared. Depth of understanding is not how many concepts you can define; it is how many situations you can run a concept through.

Second: the substrate of learning is experience, and experience can be simulated. A primitive is not installed by hearing it described — it is installed by running it, repeatedly, until it becomes automatic, the way a programmer eventually “sees” the code execute in their head without paper. And the deep, almost unsettling truth is that the mind does not distinguish between a real situation and a fully-occupied simulated one. Give a person a role — make them, for one hour, the city’s crisis manager, the prosecutor, the failing startup’s founder, the physicist proving a theorem — and they will run the same internal operations, feel the same pulls, make the same characteristic errors as they would in the real thing. The role is the only thing that is required. This is why the entire apparatus of human education leaving the richest possible substrate untouched — the young mind’s capacity to inhabit simulated situations and run real cognition inside them — is one of the great unforced errors of our civilization. Simulation is not a lesser version of reality for the purposes of installing primitives. For the purposes of installing primitives, simulation is reality.

Third: there are two great kinds of primitive, and a culture that develops only one produces cripples. The operations that let you decompose a problem and solve it — call this the IQ axis — are real, trainable, and gloriously underexploited. But there is a second axis, the operations that let you read what you and others feel, hold a boundary, enter a role, and transmit your understanding into another mind — the EQ axis — and without it the IQ axis is sealed in a jar. The most common tragedy of the gifted is not a deficit of intelligence but a deficit of transmission: an extraordinary generator with no cable to the grid. And the popular caricature of “low EQ” as merely being an asshole misses half the failure mode. The elegant formulation is this: people-pleasing means you do not understand yourself; being an asshole means you do not understand others; a boundary means you understand both. Emotional intelligence is not softness. It is the family of primitives that lets every other primitive reach people.

With those three walls standing, here is the AGI stake, because this is an Intelligence Strategy article and the intelligence lens changes everything it touches. A large language model is, quite literally, a generator — a machine that takes a context and a base of compressed knowledge and produces an output, token by token. We have spent seventy years and trillions of dollars discovering how to build a generator in silicon, and the operations we found we had to engineer into it — attention, composition, in-context inference, self-correction, search — are the same primitives this article says we should be installing in children. The pedagogy of primitives and the architecture of intelligence are not two subjects. They are one subject seen from two sides. When intelligence becomes cheap, continuous, and agentic — purchasable by the token — the scarce thing is no longer the generator. The scarce thing is the human who knows which primitives to run, in which order, on which problem, toward which end. The storage theory of mind was always wrong. In the agentic era it is also suicidal, because storage is precisely the thing the machines now do for free.

What follows is the full map: the six families of the Generative Loop, the 31 primitives pre-listed, each one then expanded with its operation and its trigger, the turn where the whole structure meets AGI, and a phased plan for the institution that should have been built around this all along — the school.

The Six Families — The Generative Loop

The Primitive Stack is organized by what the operation does to a mental state, and those functions form a loop. You can enter it anywhere and run it in any order — real thinking spirals, recurses, and jumps — but the families are genuinely distinct phases, and naming them is half the power.

A · DISCERNMENT — where to point the mind. Before any thinking happens, attention must be aimed. Discernment is the family of operations that decide what is worth thinking about at all — what you don’t know, what matters, what has value, what is excellent, what comes first. A mind weak in Discernment is busy and useless: it works hard on the wrong things. This is the most undertaught family in existence, because schools pre-select the problem for the student and thereby amputate the primitive that chooses problems.

B · MODELING — the structure of reality. Once aimed, the mind must build a model of how the thing works. Modeling is the family that takes things apart, puts them together, traces cause to effect, links the new to the known, and abstracts a reusable framework from a mess of particulars. This is the family most people mean by “understanding,” and it is far more constructive — more like building — than the passive word “understanding” suggests.

C · INQUIRY — the question–test–correct engine. Modeling without Inquiry calcifies into dogma. Inquiry is the family that generates the right question, conjectures a testable answer, runs the cheap experiment, reads the error as information, watches its own thinking, and stress-tests its own conclusions. It is the scientific method internalized as a set of personal reflexes — and it is the engine of all self-correction.

D · CREATION — producing new thought. Inquiry refines what exists; Creation makes what did not. This is the family of originality, depth, elegance, contrast, scenarios, and formulation — the operations that recombine distant concepts, refuse the surface, compress the complex into the simple-but-not-primitive, and give an inner intuition a shape that can survive in another mind. It is the most romanticized family and the most teachable, once you stop treating creativity as a personality trait and start treating it as a set of composable moves.

E · ENACTMENT — turning thought into action in the world. Thought that never reaches action is a closed loop that warms no room. Enactment is the family that applies a principle to a concrete situation, finds the highest-leverage move, commits to a decision under uncertainty, models other people’s viewpoints, transfers a capability into a new domain, and steps fully into a role. It is the bridge from the mental field to the physical world.

F · AGENCY — the self that runs the primitives. Beneath all five phases sits the substrate: the self that wants, fears, feels, persists, and integrates. Agency is the family of motivation, emotion, boundaries, courage, identity, and life-strategy — the operations that decide whether the loop runs at all, with what fuel, through what fear, toward what life. A flawless cognitive engine with no Agency substrate is a Ferrari with no driver and no road. This is the family that the IQ-obsessed forget, and it is the one that determines whether any of the rest is ever used.

The Catalogue — All 31 Primitives

Before the full expansion, here is the entire payload on one scroll. Each primitive is one irreducible operation; the families are the phases of the Generative Loop.

A · DISCERNMENT
1. Ignorance — locate exactly where your knowing breaks.
2. Relevance — find why this matters, and to whom.
3. Value — weigh worth, impact, and the moral cost of an idea or thing.
4. Quality / Taste — perceive why something is excellent.
5. Priority — rank what matters most, now.

B · MODELING
6. Mechanism & Consequence — model how it works inside, then run it forward.
7. Decompose ⇄ Compose — break a whole into parts; assemble parts into a system.
8. Connection & Analogy — link the new to the known; map structure across domains.
9. Framework — build a reusable structure that interprets many cases.

C · INQUIRY
10. Question & Hypothesis — frame the opening question; conjecture a testable answer.
11. Experiment & Feedback — test in the small; read the error as information.
12. Metacognition — watch and steer your own thinking.
13. Critique — stress-test an idea to strengthen, not destroy, it.

D · CREATION
14. Originality — recombine distant concepts into the new.
15. Depth — refuse the surface; reach the real structure.
16. Elegance — simplify to the essence without losing it.
17. Contrast — clarify a concept by setting it against its opposite and its false twin.
18. Scenarios — branch into multiple possible futures.
19. Formulation — turn an inner intuition into transmissible language.

E · ENACTMENT
20. Application & Practicality — put a principle into a concrete move.
21. Efficiency / Leverage — find the highest-impact move for the least cost.
22. Decision — convert deliberation into a committed choice.
23. Perspective — act with other people’s viewpoints modeled.
24. Transfer — deploy a capability in a new domain.
25. Role — enter a defined way of acting and decide from inside it.

F · AGENCY
26. Motivation — find the personal stake that fuels the work.
27. Emotion — interpret what you feel and what it protects.
28. Boundary — protect your integrity without breaking the relationship.
29. Courage — enter the role or action before you feel ready.
30. Identity — integrate a capability into who you are.
31. Strategy — point the whole loop at a life direction.

The Primitives

Each primitive below uses the same shape: an essence, the operation (input → output) that defines it, why it matters (its leverage and its failure mode), and the trigger questions that fire it. The triggers are the practical payload: they are the literal sentences a person — or a teacher, or a curriculum — uses to run the primitive on demand. A primitive you cannot trigger is a primitive you do not own.

A · DISCERNMENT

1. Ignorance

The most important operation almost no one is taught: knowing exactly where your knowing ends.
Operation: a topic + a felt vagueness → split it into parts and locate the precise seam where understanding breaks → a sharp map of what to learn next.
Why it matters: the weak learner says “I don’t get it” and stalls; the strong learner says “the step from A to B is where it breaks” and moves. The failure mode is comfortable fog — mistaking familiarity for understanding, which is why people who can define a concept are so often unable to use it.
Trigger: Which exact step can’t I do? Which concept do I only know dictionary-deep? Where would I fail if I had to teach this to someone tonight?

2. Relevance

The brain refuses to fund what has no context; Relevance is the operation that wires the funding.
Operation: a piece of knowledge → trace it to a life situation, a real decision, and the cost of not knowing it → a reason worth spending energy on.
Why it matters: most “laziness” in learning is not a character defect but a correct refusal to invest in something that has been stripped of all context. The failure mode is inert knowledge — material learned for the test and evaporated by Friday because it was never connected to anything the person actually does.
Trigger: Where does this show up in a real life? What decision gets better if I understand it? What mistake does someone make who doesn’t?

3. Value

Not everything interesting is important, and not everything new is good. Value is the operation that tells them apart.
Operation: an idea or thing → weigh its worth, its impact, its cost, and the harm it does, against whom it serves → a judgment of whether it is worth it, and whether it is good.
Why it matters: this is where capability becomes conscience. The failure mode is the brilliant amoral move — a solution that is efficient and elegant and quietly destructive, because the person ran every primitive except this one.
Trigger: For whom is this valuable? What pain does it remove? Is the value larger than the cost — and larger than the damage it does on the way?

4. Quality / Taste

You cannot make something good if you cannot perceive why good things are good.
Operation: an example output → compare it against criteria of excellence and find the gap to the ideal → a standard, a felt sense of why this is good and that is mediocre.
Why it matters: taste is the internal gradient that improvement climbs; without it, a person produces things but cannot make them better, because “better” is invisible to them. The failure mode is competent mediocrity — endless output with no ascent.
Trigger: Why is this good? What exactly raises its quality? What would make it better? What separates the average version from the excellent one?

5. Priority

In a complex world the scarce resource is not information but attention; Priority is the operation that allocates it.
Operation: a list of options under finite attention → rank by value × urgency × what-it-unlocks → an order of what to do now.
Why it matters: prioritization is simultaneously a strategic, practical, and moral skill — what you choose to attend to is what you value. The failure mode is busy negligence — the conviction that something “doesn’t matter,” which is the single largest brake on human and civilizational progress.
Trigger: What is the most important thing right now? What unlocks the other things? What is just noise dressed as urgency?

B · MODELING

6. Mechanism & Consequence

To understand a thing is to hold a model of how it works — and to run that model forward.
Operation: a phenomenon → build its internal causal model (what acts on what), then run it forward to project effects → a working model that explains the present and predicts the next state.
Why it matters: mechanism is the difference between knowing that something happens and knowing why, which is the difference between memorizing and engineering. The failure mode is surface correlation — narrating what happens without the causal spine, so the model breaks the moment conditions change.
Trigger: What causes what here? Where is the main lever? What happens to the whole if I change one variable?

7. Decompose ⇄ Compose

The two-directional operation that defeats complexity: take it apart, then build it back as a system.
Operation: a complex whole (or a pile of parts) → break it into sub-problems and dependencies / assemble parts into a working architecture → a solvable structure, or a built system.
Why it matters: most paralysis in front of a hard problem is the failure to see that the fog is actually a set of smaller, nameable pieces. This is the core primitive of programming, engineering, institutions, and strategy alike. The failure mode is the undifferentiated lump — treating a composite problem as one indivisible difficulty.
Trigger: What is this made of? What must I solve first? How do the pieces fit into a working whole — and where would that whole fail?

8. Connection & Analogy

Intelligence is not the number of things you know; it is the density of links between them.
Operation: a new concept → link it to what you already know, and map structure from a distant domain onto it → a denser knowledge network and a new way of seeing.
Why it matters: analogy is the engine of abstract reasoning — to say “a school should be a playground“ is to import an entire structure (experiment, role, safe failure, mastery-through-play) in four words. The failure mode is isolated facts — knowledge stored as disconnected islands that can never be retrieved when a novel situation needs them.
Trigger: What does this resemble? Where have I seen this structure before? What does the analogy reveal — and exactly where does it break?

9. Framework

The opposite of a one-time insight: a structure you can run many situations through.
Operation: a recurring kind of problem → extract its stable dimensions and their relations → a reusable structure that interprets many cases.
Why it matters: a framework is a concept-generator industrialized — generator itself, or a business-model canvas, or democracy — and the discipline of pushing arbitrary situations through a framework deepens the framework and gives it power. The failure mode is framework worship — applying a structure long after reality has stopped fitting it.
Trigger: What does a situation of this type always contain? Does this apply to more than one case? What questions does the framework force me to ask?

C · INQUIRY

10. Question & Hypothesis

The quality of a mind is bounded by the quality of the questions it can ask itself.
Operation: a vagueness or a goal → frame the question that opens the next level, then conjecture a testable answer → a productive question plus a candidate explanation.
Why it matters: a weak question — “what should I learn?” — produces a weak search; a strong question — “what mental operation must I install to solve this class of problem repeatedly?” — reorganizes the whole inquiry. The failure mode is the dead question — asking for a fact when the situation needed a mechanism, a value, or a strategy.
Trigger: What question would help me most right now? Am I asking for a fact, a mechanism, a value, or a move? What would an expert ask here?

11. Experiment & Feedback

An error is not a verdict on your worth; it is a sensor reading. Read it.
Operation: a hypothesis → test it in the small, observe the deviation from what you expected → error converted into information, and an improved next attempt.
Why it matters: this is the loop that turns flailing into learning; the person who runs it treats every failure as a data point rather than a wound. The failure mode is error as shame — the school-trained reflex to hide and fear mistakes, which severs the single most valuable feedback channel a mind has.
Trigger: How do I test this cheaply? What exactly didn’t work? Which assumption was wrong? What is the smarter next attempt?

12. Metacognition

The operation of watching your own thinking as if it were an object on a table.
Operation: your own thinking-in-progress → observe it from outside; catch yourself guessing, avoiding the hard part, or rushing to a conclusion → more accurate thinking.
Why it matters: metacognition is the conductor that decides which other primitive should be playing; without it, the mind runs on autopilot and never notices it has skipped a step. The failure mode is unwatched cognition — confusing the feeling of certainty with the fact of proof.
Trigger: How am I thinking right now? What am I assuming without checking? Am I mistaking confidence for evidence? Where did I skip a step?

13. Critique

Not cynicism — the disciplined search for the weak joint, in service of strengthening it.
Operation: a claim or idea → surface its assumptions, build the strongest counter-argument, find the load-bearing weakness → a stronger version of the idea.
Why it matters: good critique improves; it asks “where is this naive, overstated, untested, or dangerous?” and then repairs rather than discards. The failure mode splits two ways — defensive blindness (unable to attack your own idea) and destructive cynicism (attacking without rebuilding).
Trigger: What is weakest here? What am I lying to myself about? What is the best objection — and how would I answer it without throwing the idea away?

D · CREATION

14. Originality

Originality is rarely creation from nothing; it is collision between things kept apart.
Operation: two or more distant concepts → combine them under a new tension or in a new context → an original hypothesis or framing.
Why it matters: the move “what happens if I join education and simulation, or school and playground, or mind and generator?” is the literal mechanism of novelty. The failure mode is recombination of the near — only ever combining adjacent ideas, which produces variation but never surprise.
Trigger: What happens if I join A and B? Where does this pattern exist in a totally unrelated field? What combination here has no one tried?

15. Depth

The refusal to accept the surface as the answer.
Operation: a surface opinion → ask what produces it, what hidden assumption it rests on, what structure manufactures it → the real problem underneath the visible one.
Why it matters: a deep mind does not say “school is bad”; it asks what kind of consciousness school produces, what relationship to not-knowing it builds, what obedience is encoded in its very form. The failure mode is the plausible shallow — an answer that sounds right and stops exactly one layer above the truth.
Trigger: What is the real problem under the visible one? What does everyone assume without examining? What would have to be true for this to make sense?

16. Elegance

To compress a complex reality into a simple form without amputating its essence.
Operation: a complex situation → strip away everything that is not load-bearing → a simple, transmissible, non-primitive formulation.
Why it matters: elegance is what makes a truth portable — “people-pleasing means you don’t understand yourself; being an asshole means you don’t understand others” survives in a mind precisely because it is compressed without being dumbed down. The failure mode is false simplicity — cutting so deep you remove the truth along with the complexity.
Trigger: What is the simplest version that is still true? What is the core? Can I say it in one sentence without losing the depth?

17. Contrast

Many concepts only become clear the moment you set them against what they are not.
Operation: a concept → place it against its opposite and its most common false twin → a sharper concept with a defensible boundary.
Why it matters: real learning versus memorizing; understanding versus definition; a boundary versus people-pleasing; elegance versus mere simplicity — each pair teaches by opposition. The failure mode is the blurred concept — a word used confidently while quietly overlapping with three other words.
Trigger: What is this NOT? What is it most often confused with? How do I tell the real version from the counterfeit?

18. Scenarios

The future is not one line; it is a branching set, and the strong mind holds several branches at once.
Operation: a present situation + its key uncertainties → branch into several plausible futures and the triggers that select them → a map of futures to prepare for.
Why it matters: scenario-thinking is how a mind escapes the trap of a single predicted future — “if we put AI into school, it could liberate learning or outsource all thinking; which fork, and what selects it?” The failure mode is single-future tunnel vision — planning as if the one imagined outcome were certain.
Trigger: What could happen? What are the three realistic branches? What would each one mean? What should I be ready for either way?

19. Formulation

An intuition you cannot put into words is an asset you cannot bank, lead with, or transmit.
Operation: an inner intuition → give it a concept, a structure, and an example → a thought that survives intact inside someone else’s head.
Why it matters: formulation is where private genius becomes public influence — it is the primitive that decides whether your insight changes anyone or dies with you, and it is decisive for teaching, leadership, science, and founding. The failure mode is the mute intuition — being right and unable to make anyone see it.
Trigger: What am I actually trying to say? What word is missing? What example would show it? How do I phrase it so it survives in another mind?

E · ENACTMENT

20. Application & Practicality

A principle that never touches a concrete situation is decoration.
Operation: a principle → drop it into a specific situation under real constraints → a usable move or a testable prototype.
Why it matters: to understand incentives is to be able to find them in a school, a firm, a government, a family, and your own life — application is the proof that a concept is owned and not merely recited. The failure mode is the floating abstraction — deep ideas that never descend into a single thing you could do tomorrow.
Trigger: What would this look like in practice? What is the first small experiment? Who does it, and how would we know it worked?

21. Efficiency / Leverage

Not “do more” but “find the one move that moves the most.”
Operation: a goal + finite resources → search for the point of maximum impact at minimum cost → a high-leverage move.
Why it matters: leverage thinking asks where the lever is rather than how hard to push — often the answer is to change the form of the work rather than add to its quantity. The failure mode is effort theater — heroic exertion on a low-leverage point.
Trigger: Where is the biggest lever? What can I remove entirely? How do I get eighty percent of the result with twenty percent of the effort?

22. Decision

Thinking that never closes into a choice is an infinite loop dressed as diligence.
Operation: options + criteria → weigh risk, reversibility, and preference → a committed choice.
Why it matters: the decision primitive is what converts deliberation into motion, and its quality depends on distinguishing reversible bets (decide fast) from irreversible ones (decide slow). The failure mode is analysis paralysis — endless reflection used as a sophisticated way to avoid the discomfort of committing.
Trigger: What are the real options? On what criteria am I choosing? What is the biggest risk? Which parts are reversible and which are not?

23. Perspective

A mind trapped in its own viewpoint is a mind that will be surprised by half of reality.
Operation: a problem → re-run it through other actors’ motivations, fears, and information → a richer, less self-trapped understanding.
Why it matters: seeing a situation through the child, the teacher, the parent, the state, the employer, the outsider is how you find the moves your own position made invisible. The failure mode is egocentric modeling — assuming everyone shares your information and incentives.
Trigger: How does someone else see this? What is rational from where they stand? What do they know that I don’t? What are they afraid of?

24. Transfer

A skill that only works where you first learned it is a skill you barely have.
Operation: a capability learned in one domain → extract its deep, domain-independent principle and adapt it to a new field → a capability that works outside its origin.
Why it matters: programming teaches decomposition, abstraction, modularity, and testing — and the person who can transfer those into management, writing, or strategy has multiplied one course into ten. The failure mode is context-locked skill — knowing the technique but not the principle, so it never travels.
Trigger: What general principle did I actually learn here? Where else does this exact pattern hold? What part is specific and what part is universal?

25. Role

The most powerful learning instrument we own and the most neglected: become someone, and decide from inside them.
Operation: a situation → enter a defined role and make decisions from inside its responsibility → lived experience of an identity in action.
Why it matters: a child who plays the scientist, the mayor, the founder, the judge does not learn facts about those roles — they install the role’s posture, its kind of decision, its weight of responsibility, and because the mind does not distinguish a fully-occupied simulation from reality, the learning is real. The failure mode is spectator learning — watching a role described instead of inhabiting it.
Trigger: How would a scientist / founder / mayor act here? What responsibility does this role carry? What do I learn only by stepping inside it?

F · AGENCY

26. Motivation

The loop does not run on command; it runs on fuel, and Motivation is the operation that finds the fuel.
Operation: a topic → find the personal stake in it — a tension, a fascination, a future self → the energy to actually engage.
Why it matters: a person learns fastest when the material stops being a foreign object and becomes their question, tied to their life and their curiosity. The failure mode is extrinsic-only drive — running on grades and fear, which collapses the instant the external pressure is removed.
Trigger: What is alive in this for me? When would I actually want this ability? What genuinely interests me underneath the assignment?

27. Emotion

Feelings are not noise in the signal; they are information about the relation between you and the situation.
Operation: a feeling → read what it protects and what need it signals → emotional understanding and a response adequate to the situation.
Why it matters: emotions report on value, threat, and need — to interpret them is the foundation of self-knowledge and of every relationship, and a mind that cannot read its own feelings cannot read anyone else’s. The failure mode splits into alexithymia (not understanding yourself) and projection (not understanding others).
Trigger: What am I feeling right now? What is it protecting? What need is it pointing at? Is my reaction adequate to what actually happened?

28. Boundary

To protect your own reality without destroying the other person’s — the load-bearing operation of emotional intelligence.
Operation: your own need under outside pressure → hold your line while keeping the relationship intact → a healthy self-definition.
Why it matters: people-pleasing means you don’t understand yourself; being an asshole means you don’t understand others; a boundary means you understand both — boundaries are where self-knowledge and other-knowledge meet. The failure mode is the two-sided collapse: being steamrolled or being the steamroller.
Trigger: What is genuinely unacceptable to me? Where am I letting myself be steamrolled? How do I say it firmly without making it an attack?

29. Courage

Most capability is gated not by ability but by the willingness to enter before you feel ready.
Operation: a challenge + the fear it produces → take the smallest safe step into the role or action before feeling competent → an expanded capacity to act.
Why it matters: speaking up, leading, arguing, creating, admitting you don’t know — these are learned only by entering, and the role almost always precedes the confidence to occupy it. The failure mode is the readiness trap — waiting to feel ready for an experience that only readiness-through-doing can ever provide.
Trigger: What am I afraid of here? What is the smallest step in? What will I learn that is available only by entering?

30. Identity

A skill becomes permanent only when it stops being something you do and becomes someone you are.
Operation: a repeated, mastered experience → name it inwardly as a role you now hold → a new, load-bearing part of who you are.
Why it matters: the deep change is the shift from “I am learning to write” to “I am someone who can give thoughts form” — identity is the internal permission to use a capability without hesitation. The failure mode is the impostor gap — possessing a skill while withholding from yourself the right to claim it.
Trigger: What kind of person does this make me? When have I already done it? How would I act if this were simply my nature?

31. Strategy

The operation that points the entire loop at a life — not “what can I do?” but “where is all of this going?”
Operation: your goals, strengths, gaps, and values → align the whole Generative Loop toward a direction → a personal development strategy.
Why it matters: education has a point only if it helps a person steer their life — choosing which abilities to build, which experiences will grow them, which environments will force them upward. The failure mode is drifted competence — accumulating skills with no direction, the over-specialized expert who is exquisitely sharp and entirely lost.
Trigger: Where am I trying to get to? Which abilities am I missing? Which experiences would move me most? What kind of work or life would force me to grow?

How AGI Changes the Game

Every Intelligence Strategy framework must be struck against the intelligence lens, and the Primitive Stack is no exception. When intelligence becomes cheap, continuous, and agentic, the value of each primitive does not stay fixed — it re-prices, and the re-pricing is the whole strategic story.

The core inversion is this. For the first time, the generator is not the scarce asset. A human mind used to be the only available machine that could take a context and a knowledge base and produce a structured output. That monopoly is over. A large language model is a generator you can rent by the token, and it runs many of the 31 primitives — Decompose, Mechanism, Connection, Formulation, Scenarios, Critique — faster and more tirelessly than any person. The naive conclusion is that human primitives are now worthless. The correct conclusion is the opposite, and sharper: when the running of primitives is commoditized, the scarce skill becomes knowing which primitive to run, in what order, on what problem, toward what end — and that meta-skill is itself made of primitives, the ones machines run worst: Value, Priority, Taste, Boundary, Identity, Strategy. The human edge migrates up the stack, from MODELING and CREATION toward DISCERNMENT and AGENCY.

Read as a set of shifts:

1. From storage to selection. From the educated person as the one who has memorized the most to the one who can discern what is worth attending to — because the warehouse is now free and infinite. Discernment was always the real skill; AGI has merely made that undeniable.

2. From answering questions to asking them. From a premium on producing the answer to a premium on framing the question (Primitive 10) — the machine answers superbly and questions poorly, so the human who asks the sharp question commands the machine that answers it.

3. From having ideas to judging them. From idea-generation as the bottleneck to idea-selection as the bottleneck — when a generator can produce a hundred plausible options, Value, Quality, and Critique become the rate-limiting primitives, not Originality.

4. From private cognition to externalized, editable cognition. From thinking trapped invisibly in one head to thinking rendered immediately as an artifact a machine can extend and a person can inspect — which raises the return on Formulation and Metacognition, the primitives that govern that boundary.

5. From individual generator to orchestrated generators. From the lone mind solving the problem to the human orchestrating a swarm of machine generators — which makes Decompose-⇄-Compose, Role, and Strategy the operations of leverage, because directing many generators is a composition problem.

6. From transmission as bottleneck to transmission as multiplier. From EQ as a “soft skill” to EQ as the primitive family that decides whether your amplified output reaches and moves other humans — in a world where everyone can generate, Perspective, Boundary, and Formulation are the difference between noise and influence.

The trade-off that runs through every one of these shifts is the same, and it must be named: a generator this powerful can install primitives or atrophy them. The same AI that could let a child run a thousand simulations, inhabit a hundred roles, and receive instant feedback on every experiment can also let that child outsource the primitive entirely and never install it — a passive consumer of generated answers whose own generators never switch on. The technology is neutral; the pedagogy is not. Which fork we take is the central educational decision of the agentic era — and it is itself an instance of Primitive 18, Scenarios, run at civilizational scale.

The Action Plan: From Curriculum to Primitive Install

If the mind is a stack of primitives and education is their installation, then the school we have is built around the wrong noun. It is organized to transmit content when it should be organized to install operations. The content it transmits is now free; the operations it neglects are now the entire value. Here is the phased redesign — the Primitive Curriculum.

Phase 1 — Re-found the school on experience and role

The substrate of installation is experience, and the richest available substrate is the simulated role. The first move is structural: make the school a playground for life rather than a delivery system for facts.

• Step 1 — Convert subjects into situations. Every topic is re-expressed as a situation a student occupies in a role: history as a chamber of political decisions made under the real constraints of the period; physics as a proving workshop where equations are derived and chained, not memorized; civics as a city in crisis that the students must govern.

• Step 2 — Install through repetition, not exposition. Each situation is run until the target primitive becomes automatic — the way a programmer eventually sees the code execute without paper. The unit of progress is “can the student trigger the primitive on demand?”, not “was the student exposed to the content?”

• Deliverable: the Situation Library — a bank of role-based simulations, each tagged with the primitives it installs.

Phase 2 — Make the primitives explicit and trainable

A primitive named is a primitive that can be practiced; a primitive left implicit is left to chance.

• Step 1 — Teach the triggers as first-class content. Students learn the literal trigger-questions of each primitive — “which exact step can’t I do?”, “where is the biggest lever?”, “what is this NOT?” — as the actual curriculum, the way one learns multiplication tables. The triggers are the moves; the moves are the lesson.

• Step 2 — Train metacognition as the conductor. Students are taught to name which primitive they are running — “I’m decomposing now,” “I’m critiquing now,” “I skipped Relevance” — so the loop becomes visible and steerable rather than automatic and invisible.

• Deliverable: the Primitive Logbook — a record in which each student tracks which primitives they can reliably trigger, and which remain fog.

Phase 3 — Build the two axes together, never one alone

The IQ axis without the EQ axis produces sealed genius; the redesign refuses the split.

• Step 1 — Weight the Agency family equally. Motivation, Emotion, Boundary, Courage, Identity, and Strategy are taught as operations, not as the vague pastoral residue left over after “real” subjects. A student who cannot hold a boundary or read an emotion is treated as having an un-installed primitive, not a personality.

• Step 2 — Make transmission a graded output. Because EQ is the family that lets every other primitive reach people, students are assessed on Formulation and Perspective directly: can you make another mind see what you see?

• Deliverable: the Whole-Loop Profile — a per-student map across all six families, replacing the single ranked grade with a picture of which generators are installed and which are not.

Phase 4 — Wire in the machine generators deliberately

The agentic era arrives in the classroom whether we plan for it or not; the only choice is whether it installs primitives or atrophies them.

• Step 1 — Use AI to multiply experience, not to replace it. Machine generators run the thousand simulations, play the counterpart roles, and deliver instant feedback on every experiment — radically expanding the substrate of lived situations a young mind can run through.

• Step 2 — Forbid the silent outsource. The non-negotiable rule: the machine may expand a student’s loop but never run it for them unobserved. Every AI-assisted task is paired with the metacognitive demand to name which primitive the student themselves executed — so the human generators switch on rather than going dark.

• Deliverable: the Augmentation Protocol — an explicit standard for which primitives a student must always run unaided, and which the machine may amplify.

The storage theory of mind gave us schools that fill warehouses, exams that measure the warehouse, and a civilization that mistook the size of the warehouse for the power of the mind. It was a defensible error in an age when storage was expensive and generators were rare. It is an indefensible one now, in an age when storage is free and the only scarce thing left is a human who knows which generators to run and toward what end.

The mind is not a database. It is a generator — a stack of 31 primitives, composed into a single Generative Loop, installed through experience and role, and pointed, by the Agency family that powers it, at a life. Build the school around that noun, and you do not produce people who have memorized the world. You produce people who can generate it.

The first faculty is Comprehension: the ability to see how the system actually works—what the existing law already says, where it contradicts itself, which statutes are dead, who really benefits. Today this faculty barely exists; legislators vote on text they have not read and cannot, structurally, find time to read. An agent reads all of it, continuously, and turns the opaque corpus of accumulated law into a queryable map.

The second faculty is Significance: the discipline of deciding which problems are even worth a law. Legislative attention is the scarcest resource in a republic, and it is allocated by noise—by whichever crisis trends, whichever lobby shouts loudest. An agent can triage a thousand candidate problems by reach, severity, and reversibility, turning a politics of reaction into a politics of deliberate prioritisation.

The third faculty is Tractability: the sober estimate of how hard a problem is to actually move. Most political energy is spent on problems that look urgent but are structurally immovable, while tractable wins go unnoticed. An agent can model expected effect size against implementation difficulty, separating the problems a law can solve from the ones it will only perform solving.

The fourth faculty is Diffusion: the capacity to learn from everyone who already tried. The fifty states and the hundred-ninety countries are a vast, running experiment, and almost none of that evidence reaches the drafter in time. An agent mines the entire global record of policy—what spread, what worked, what backfired—and delivers proven templates instead of blank pages.

The fifth faculty is Evidence: the loyalty to what the research actually shows rather than what the talking point asserts. The evidence base is enormous and growing, and it is almost entirely unscanned by the people writing law. An agent grounds every claim in the studies, the trials, and the data—and, critically, supplies that grounding without a client behind it.

The sixth faculty is Simulation: the power to test a law before it is binding. We ship software behind a staging environment and a rollback button; we ship law to a continent on a floor vote and a hope. An agent war-games legislation against a synthetic population, surfacing the second- and third-order effects—the cobra-breeders, the gaming, the perverse incentives—in silico, before they hit reality.

The seventh faculty is Composition: the act of turning settled intent into precise statutory text. This is the one task already visibly migrating to machines, from a city ordinance drafted by a chatbot to a national drafting assistant trained on a million sections of law. Done well, it collapses the cost of writing good law; done carelessly, it floods the system with bad law faster than ever.

The eighth faculty is Constituent Sensing: the ability to hear what the public actually needs, directly and at scale, rather than through the filter of whoever can manufacture the loudest voice. Today a representative’s sense of the public is a handful of town halls and a flood of form letters; when millions of comments arrive, the genuine signal drowns. An agent listens to all of it, strips out the astroturf, and renders the real distribution of need.

The ninth faculty is Deliberation: the discipline of forcing a proposal to survive its strongest objections before it becomes law. Legislatures vote under time pressure and tribal reflex, rarely steelmanning the other side or naming who pays. An agent cross-examines every bill—generating the best case against, the trade-offs, and the role-reversal test—so the decision rests on public reasons, not on whoever held the floor.

The tenth faculty is Oversight: the loop that learns whether a law actually worked. Most legislation is passed once and never revisited, accumulating as dead statute no one tests. An agent measures every law against its own stated goals, flags failure early, and triggers the revision or repeal that turns lawmaking from a one-way act into a system that learns.

This article is a field guide to the ten capabilities of the Augmented Legislator—the Legislative Intelligence Stack, where Constituent Sensing brackets the front of the cycle and Oversight the back, with Deliberation standing between knowing and writing. Each capability is treated identically: a precise Definition, its Place in lawmaking in five aspects, the twelve principles that make it powerful, the three patterns by which it operates, the key mechanisms with real working examples, the way agents change the game, the four principles of that shift, and the honest advantages and disadvantages. The article closes with a phased Action plan for building the Stack inside a real legislature without surrendering the one thing that must remain human: the vote.

Summary

1) Comprehension

What it is — The faculty of seeing the existing system as it really is: the full corpus of law, its contradictions, its dead letters, its true beneficiaries.
How it works — Continuous reading and structural mapping of statutes, precedents, and proposed text into a queryable model.
Why it matters — You cannot reform a system you cannot see; comprehension is the precondition for every other faculty.
Failure mode — Voting blind: passing text no human has read or understood, captured by whoever summarises it.

2) Significance

What it is — The triage faculty: deciding which problems are meaningful enough to deserve scarce legislative attention.
How it works — Scoring candidate problems by reach, severity, urgency, and reversibility into an explicit priority order.
Why it matters — Attention is the binding constraint of a republic; misallocating it wastes the whole machine.
Failure mode — Government by trending crisis: loud problems crowd out large ones.

3) Tractability

What it is — The realism faculty: estimating how hard a problem is to actually move with a law.
How it works — Modelling expected effect size against implementation difficulty, cost, and resistance.
Why it matters — Effort spent on immovable problems is the largest hidden waste in politics.
Failure mode — Performative legislation: passing laws that look like solutions but cannot bite.

4) Diffusion

What it is — The learning faculty: mining other jurisdictions for policies that already worked.
How it works — Scanning the global record of adoption, outcomes, and failures to surface proven templates.
Why it matters — Most problems have been solved somewhere; reinvention is pure waste.
Failure mode — Parochial blindness: drafting from scratch while the answer sits in another statehouse.

5) Evidence

What it is — The grounding faculty: tying decisions to what research and data actually show.
How it works — Retrieving, weighing, and citing studies, trials, and evaluations against each claim.
Why it matters — Without evidence, law is narrative; with it, law can be corrected.
Failure mode — Lobbyist epistemics: the best-funded interest supplies the “facts.”

6) Simulation

What it is — The foresight faculty: testing a law against a model of the world before it is binding.
How it works — War-gaming policy on synthetic populations and economic models to expose second-order effects.
Why it matters — Unintended consequences are where good intentions go to die.
Failure mode — Shipping to 330 million people with zero unit tests.

7) Composition

What it is — The drafting faculty: converting settled intent into precise, conflict-free statutory text.
How it works — Generating and red-lining legal language grounded in the existing corpus.
Why it matters — The gap between intent and text is where loopholes and litigation live.
Failure mode — Legislative spam: cheap drafting that floods the system with volume, not law.

8) Constituent Sensing

What it is — The input faculty: hearing what citizens actually need, at scale, beneath the manufactured noise.
How it works — Collecting, deduplicating, and classifying public input while filtering astroturf and fraud.
Why it matters — A representative who cannot hear the represented governs blind to them.
Failure mode — Mistaking the loudest manufactured campaign for the public will.

9) Deliberation

What it is — The reasoning faculty: stress-testing a decision against its strongest objections.
How it works — Generating the opposing case, the trade-offs, and the role-reversal test.
Why it matters — A law unexamined by its best critics is a law waiting to fail.
Failure mode — Tribal reflex: passing on “our side” rather than on public reasons.

10) Oversight

What it is — The feedback faculty: learning whether a law actually worked after passage.
How it works — Measuring real outcomes against stated goals and triggering revision or repeal.
Why it matters — Without a feedback loop, laws accumulate as dead, unexamined sediment.
Failure mode — Ghost laws: passed once and never revisited.

The Capabilities

1) Comprehension

Definition

Comprehension is the faculty of accurately seeing the system a legislator proposes to change—the full body of existing law, its internal contradictions, its obsolete provisions, and its real-world beneficiaries—before touching it.

It functions as the legislature’s situational awareness layer: the precondition that makes every downstream faculty possible, because no problem can be triaged, no law simulated, and no text drafted against a system that is invisible to the person governing it.

Place in lawmaking: 5 aspects

1. The precondition for legitimacy

   • A vote on unread text is a vote without consent of the mind that casts it.

   • Comprehension is what converts a signature into an actual decision.

2. The map of the existing corpus

   • Statute is accreted over centuries; no single mind holds it.

   • Comprehension turns that sediment into a navigable structure.

3. The contradiction detector

   • New law collides with old law in ways drafters rarely foresee.

   • Comprehension surfaces conflicts before they become litigation.

4. The dead-letter finder

   • Much law is obsolete, redundant, or never enforced.

   • Comprehension distinguishes living rules from fossils.

5. The beneficiary lens

   • Every rule moves value to someone; the question is whom.

   • Comprehension makes the distributional reality legible.

Why it works: 12 principles

1. Externalised memory — it stores the corpus outside any single overloaded staff.

2. Structural reading — it maps relationships (this section amends that one), not just words.

3. Completeness — it reads all of the text, not the fraction a human samples.

4. Cross-reference — it links proposed text to every statute it touches.

5. Provenance — it traces where language came from and who supplied it.

6. Comparability — it sets current law beside the proposed change, clause by clause.

7. Continuity — it persists across electoral cycles, immune to staff turnover.

8. Speed — it reads in minutes what once took staff weeks.

9. Searchability — any clause becomes retrievable on demand.

10. Version awareness — it tracks how text mutated across drafts and amendments.

11. Scale-invariance — a thousand-page bill is no harder to read than a one-pager.

12. Neutrality — it gives every provision equal attention, not selective focus.

Three major patterns of how it works

1. Ingest → structure → query

   • Ingest the raw corpus and the proposed text

   • Structure it into linked clauses, definitions, and cross-references

   • Expose it to natural-language interrogation

2. Compare → flag → explain

   • Compare new language against existing law

   • Flag conflicts, redundancies, and dead letters

   • Explain each flag in plain language

3. Trace → attribute → expose

   • Trace clauses to their textual origin

   • Attribute them to a source (agency, interest, model bill)

   • Expose the provenance to the legislator and the public

Key mechanisms and how they work

A. Statutory research and pruning systems

• Models that read the entire code and locate the relevant, redundant, or obsolete law.

• Example: Stanford’s RegLab built a statutory-research system that identified relevant law with 94–99% reliability; deployed with the San Francisco City Attorney, it produced an ordinance cutting more than a third of the city’s mandated reports.

B. Code-scale deregulation analysis

• Running an entire administrative code through analysis to flag what is unnecessary.

• Example: Ohio ran its roughly fifteen-million-word administrative code through an AI analysis that flagged two million words and some 900 rules for removal, putting the state on track to cut nearly a third of the code.

C. Provenance and model-legislation detection

• Computational comparison that reveals who actually wrote a bill.

• Example: The “Copy, Paste, Legislate” investigation analysed nearly a million state bills and found more than 10,000 copied almost verbatim from interest-group “model legislation,” over 2,000 of which became law.

How AI changes the game: definition

AI turns comprehension from a sampling problem into a total-coverage problem—reading the whole corpus, mapping its structure, and answering questions about it in real time—while shifting the risk from “we missed something” to “we over-trusted the summary.”

In short: the legislator can finally read everything.

Four principles of how AI changes the game

1. From sampling to totality — from reading a fraction of the text to processing all of it.

2. From text to structure — from prose pages to a linked, queryable graph of law.

3. From periodic to continuous — from a one-time read to an always-current model of the corpus.

4. From opaque to attributed — from anonymous clauses to traceable provenance.

Advantages and disadvantages

Advantages

1. Ends the absurdity of voting on unread text.

2. Surfaces conflicts and dead letters before they cause harm.

3. Exposes hidden authorship and beneficiaries.

4. Gives a small office the reading capacity of a large institution.

Disadvantages

1. A confident, wrong summary is more dangerous than an honest gap—automation bias is real.

2. Whoever tunes the comprehension model shapes what the legislator “sees.”

3. Structural maps can flatten the deliberate ambiguity that law sometimes needs.

4. Total legibility of the corpus is also a tool for those who would exploit it.

2) Significance

Definition

Significance is the faculty of deciding which problems are meaningful enough to warrant scarce legislative attention—weighing how many are affected, how severe the harm, how urgent the timing, and how reversible the damage.

It functions as the legislature’s triage layer: the discipline that allocates the single most constrained resource in a republic—the finite attention of its lawmakers—toward the problems that actually matter rather than the ones that merely shout.

Place in lawmaking: 5 aspects

1. The attention allocator

   • There are always more problems than legislative slots.

   • Significance decides what gets a hearing and what does not.

2. The severity weigher

   • Not all harms are equal; some are catastrophic, some cosmetic.

   • Significance ranks by magnitude, not volume of complaint.

3. The reach estimator

   • A problem affecting millions differs from one affecting hundreds.

   • Significance scales attention to population touched.

4. The reversibility filter

   • Irreversible harms deserve priority over recoverable ones.

   • Significance privileges the problems that cannot wait.

5. The agenda guard

   • Agendas are captured by whoever manufactures urgency.

   • Significance defends the agenda against manufactured noise.

Why it works: 12 principles

1. Comparability — it puts dissimilar harms on a common scale.

2. Proportionality — it matches attention to magnitude.

3. Explicitness — it makes the priority order visible and defensible.

4. Resistance to noise — it discounts volume in favour of severity.

5. Forward weighting — it privileges the irreversible and the compounding.

6. Coverage — it scans the whole problem space, not the trending slice.

7. Auditability — it leaves a record of why a problem was prioritised.

8. Multi-dimensionality — it weighs reach, severity, urgency, and reversibility together.

9. Counterfactual framing — it asks what happens if nothing is done at all.

10. Stakeholder breadth — it counts the silent affected, not only the vocal.

11. Recurrence sensitivity — it flags chronic problems that never spike but never resolve.

12. Revisability — priorities update as conditions change.

Three major patterns of how it works

1. Scan → score → rank

   • Scan the full landscape of candidate problems

   • Score each by reach, severity, urgency, reversibility

   • Rank into an explicit priority order

2. Aggregate → weight → triage

   • Aggregate signals of harm across data sources

   • Weight by magnitude and population

   • Triage into act / monitor / ignore

3. Compare → justify → publish

   • Compare a problem against the current agenda

   • Justify its place with explicit criteria

   • Publish the reasoning for scrutiny

Key mechanisms and how they work

A. Severity thresholds and common currencies

• Institutions already triage life-and-death allocation with explicit severity metrics.

• Example: The UK’s NICE allocates health spending against an explicit cost-per-quality-adjusted-life-year threshold, with a formal “severity modifier” that raises the bar a society will pay for the most severe conditions—a working machine for ranking meaningfulness.

B. Evaluation coverage as a significance signal

• Knowing which programs are unexamined reveals where attention is missing.

• Example: Reformers behind the U.S. evidence-based-policy movement estimate that only a small fraction of public spending is rigorously evaluated, and propose setting aside as little as 1% of program funds for evaluation—evidence that significance is currently unmeasured.

C. The ghost-law problem

• Laws passed and never revisited are significance failures by default.

• Example: Scoping reviews of ex-post legislative evaluation find that the societal impact of most laws is rarely measured after passage, leaving “ghost laws” on the books with no one asking whether they still matter.

How AI changes the game: definition

AI turns significance from an implicit, noise-driven reflex into an explicit, continuous triage—scoring a thousand candidate problems by reach and severity in the time a staffer reads one lobbyist memo—while raising the danger that whatever the model fails to count becomes invisible.

In short: prioritisation becomes deliberate, not reactive.

Four principles of how AI changes the game

1. From loudest to largest — from the problem that trends to the problem that matters.

2. From episodic to continuous — from crisis-driven attention to standing triage.

3. From implicit to explicit — from gut ranking to a defensible, published score.

4. From narrow to comprehensive — from the visible slice to the whole problem space.

Advantages and disadvantages

Advantages

1. Protects the agenda from manufactured urgency.

2. Surfaces large, quiet problems that never trend.

3. Makes prioritisation transparent and contestable.

4. Aligns scarce attention with actual magnitude of harm.

Disadvantages

1. What the model cannot quantify, it may silently de-prioritise.

2. Severity scoring embeds contestable value judgments as if neutral.

3. A triage metric, once public, becomes a target to be gamed.

4. Quantified significance can crowd out legitimate moral salience that resists numbers.

3) Tractability

Definition

Tractability is the faculty of estimating how hard a problem is to actually move—how large an effect a law can realistically produce, against how much cost, complexity, and resistance it must overcome.

It functions as the legislature’s realism layer: the discipline that separates problems a law can genuinely solve from problems a law can only perform solving, redirecting effort from the immovable to the achievable.

Place in lawmaking: 5 aspects

1. The effect-size estimator

   • Some interventions move the needle; many do not.

   • Tractability forecasts the realistic magnitude of impact.

2. The difficulty appraiser

   • Implementation, enforcement, and compliance all cost.

   • Tractability prices the friction of making a law bite.

3. The resistance map

   • Every law meets opposition proportional to whose value it moves.

   • Tractability anticipates where the law will be fought.

4. The leverage finder

   • Small, well-placed changes can outperform sweeping ones.

   • Tractability locates the high-leverage intervention point.

5. The futility filter

   • Some problems are structurally beyond a single statute.

   • Tractability flags where law is the wrong instrument.

Why it works: 12 principles

1. Expected value — it weighs impact by probability of success, not hope.

2. Cost realism — it counts implementation and enforcement, not just intent.

3. Resistance modelling — it forecasts opposition and capture.

4. Leverage focus — it seeks the minimal change with maximal effect.

5. Mechanism clarity — it demands a causal story for why a law would work.

6. Boundary honesty — it admits where law cannot reach.

7. Comparability — it ranks interventions by achievability, not ambition.

8. Path dependence — it accounts for what current structures actually permit.

9. Time horizon — it distinguishes quick wins from slow burns.

10. Reversibility of the fix — it favours interventions that can be undone if wrong.

11. Enforcement realism — it weighs whether a rule can actually be policed.

12. Coalition feasibility — it estimates whether the votes and allies exist to pass it.

Three major patterns of how it works

1. Model → estimate → discount

   • Model the causal mechanism

   • Estimate the raw effect size

   • Discount by implementation difficulty and resistance

2. Decompose → locate → target

   • Decompose a problem into movable and immovable parts

   • Locate the high-leverage component

   • Target the intervention there

3. Forecast → stress → revise

   • Forecast the expected outcome

   • Stress it against opposition and evasion

   • Revise the ambition to match what can bite

Key mechanisms and how they work

A. Effect-size evidence from trials

• A growing body of randomised trials gives realistic priors on how much an intervention moves.

• Example: The development-economics network J-PAL has run nearly a thousand randomised controlled trials across more than eighty countries, producing concrete effect sizes that tell a drafter whether a given lever historically moved the outcome at all.

B. Calibrated, low-cost interventions

• Cheap, well-targeted changes can have outsized, measurable effects.

• Example: The UK’s behavioural-insights work found that a single rewritten tax-reminder letter—telling recipients most neighbours had already paid—was estimated to raise tens of millions a year, a high-tractability win invisible to grand legislation.

C. The futility signal from backfires

• History records interventions whose tractability was misjudged and which moved the problem the wrong way.

• Example: Research on “three-strikes” sentencing found it flattened the penalty gradient so severely that eligible offenders became measurably more likely to commit violent crimes—an immovable problem made worse by a law that looked decisive.

How AI changes the game: definition

AI turns tractability from a gut feel into a modelled estimate—pulling real effect sizes from the global trial record and weighing them against implementation friction—while risking false precision that dresses guesswork as forecast.

In short: ambition gets calibrated to what can actually move.

Four principles of how AI changes the game

1. From hope to expected value — from “this should work” to “this historically moved X.”

2. From intent to friction — from the goal to the real cost of enforcing it.

3. From sweeping to leveraged — from grand gestures to minimal high-impact changes.

4. From certainty to calibrated doubt — from false confidence to honest probability.

Advantages and disadvantages

Advantages

1. Redirects effort from immovable problems to achievable ones.

2. Grounds ambition in real historical effect sizes.

3. Exposes the implementation friction politicians routinely ignore.

4. Surfaces cheap, high-leverage interventions that never make headlines.

Disadvantages

1. Effect sizes from one context transfer imperfectly to another.

2. Quantified tractability can bias toward the easily measured and against the structurally important.

3. A low-tractability score can become an excuse for inaction on hard, vital problems.

4. Modelled forecasts carry false precision that invites over-trust.

4) Diffusion

Definition

Diffusion is the faculty of learning from every jurisdiction that already faced a problem—mining the fifty states and the hundred-ninety countries for the policies that spread, the ones that worked, and the ones that backfired.

It functions as the legislature’s import layer: the mechanism that converts the world’s running policy experiment into proven templates, so a drafter starts from what already succeeded elsewhere rather than from a blank page.

Place in lawmaking: 5 aspects

1. The laboratory harvester

   • Sub-national and foreign governments are live experiments.

   • Diffusion harvests their results for reuse.

2. The template supplier

   • Most problems have a workable solution somewhere.

   • Diffusion supplies it instead of a blank draft.

3. The failure archive

   • Other jurisdictions have already made the mistakes.

   • Diffusion imports the warnings, not just the wins.

4. The implementation-detail carrier

   • The difference between success and failure is often a detail.

   • Diffusion transfers the how, not only the what.

5. The patchwork tracker

   • Reforms move unevenly across dozens of legislatures at once.

   • Diffusion keeps the moving map current.

Why it works: 12 principles

1. Reuse — it avoids reinventing solved problems.

2. Evidence of feasibility — a policy that ran elsewhere is proof of possibility.

3. Outcome transfer — it carries results, not just designs.

4. Failure avoidance — it imports others’ mistakes as warnings.

5. Detail fidelity — it transfers the implementation specifics that decide success.

6. Timeliness — it surfaces proven options before the drafting deadline.

7. Breadth — it scans more jurisdictions than any human could track.

8. Context matching — it weights examples by similarity to local conditions.

9. Adaptation over copying — it adjusts templates rather than transplanting them blind.

10. Recency — it privileges current results over stale precedent.

11. Counter-diffusion awareness — it tracks where reforms were repealed or banned, not only adopted.

12. Source diversity — it draws from many jurisdictions, avoiding single-model dependence.

Three major patterns of how it works

1. Scan → match → adapt

   • Scan the global record for analogous problems

   • Match the closest proven policy

   • Adapt it to local constraints

2. Trace → evaluate → import

   • Trace where a policy spread

   • Evaluate its measured outcomes

   • Import the version that worked

3. Detect → warn → adjust

   • Detect where a policy backfired

   • Warn the drafter of the failure mode

   • Adjust the design to avoid it

Key mechanisms and how they work

A. Cross-jurisdiction policy mining

• The “laboratories of democracy” only help if someone reads the results.

• Example: When New York implemented cordon congestion pricing it explicitly followed London’s earlier rollout, down to the implementation detail of pairing the charge with expanded bus service to absorb displaced drivers.

B. Tracking a reform across many legislatures at once

• Proven templates spread unevenly and fast across dozens of statehouses.

• Example: Right-to-repair legislation has now been introduced in all fifty U.S. states—a patchwork no single staffer can track, but exactly the moving map a diffusion agent maintains.

C. Global outcome evidence for a single policy

• The same policy run in many countries yields a distribution of outcomes to learn from.

• Example: More than forty countries have adopted sugar-sweetened-beverage taxes, leaving a documented range of consumption effects—from modest to large—for the next adopter to study before drafting.

How AI changes the game: definition

AI turns diffusion from occasional, anecdotal borrowing into systematic, continuous mining of the entire global policy record—surfacing proven templates and documented failures on demand—while risking the uncritical transplant of policies whose context does not travel.

In short: the drafter starts from what already worked.

Four principles of how AI changes the game

1. From anecdote to corpus — from a remembered example to the whole record.

2. From design to outcome — from copying a law’s text to copying its measured results.

3. From wins-only to failures-included — from cherry-picked success to honest distribution.

4. From snapshot to live map — from a one-time scan to a continuously updated tracker.

Advantages and disadvantages

Advantages

1. Eliminates the waste of reinventing solved problems.

2. Carries implementation details that decide success or failure.

3. Imports others’ mistakes as cheap warnings.

4. Keeps a live map of reforms moving across many jurisdictions.

Disadvantages

1. A policy that worked in one context can fail in another; transplant is risky.

2. Diffusion can entrench convergence and suppress local experimentation.

3. The same machinery lets interest groups spread model legislation faster, too.

4. Outcome data from abroad is uneven, lagged, and sometimes politicised.

5) Evidence

Definition

Evidence is the faculty of grounding legislative decisions in what research and data actually show—retrieving, weighing, and citing the studies, trials, and evaluations that bear on a claim, rather than the assertions supplied by whoever is in the room.

It functions as the legislature’s grounding layer: the discipline that ties law to reality and, decisively, supplies that grounding without a client—breaking the monopoly under which the best-funded interest is also the source of the “facts.”

Place in lawmaking: 5 aspects

1. The reality anchor

   • Law detached from evidence is narrative with force.

   • Evidence keeps the claim tethered to the world.

2. The subsidy replacement

   • Today, research and drafting labour is donated by lobbyists.

   • Evidence supplies the same subsidy with no donor attached.

3. The claim auditor

   • Every justification rests on an empirical premise.

   • Evidence checks whether the premise is true.

4. The uncertainty reporter

   • Honest evidence carries its own error bars.

   • Evidence states what is known and what is not.

5. The correction enabler

   • Only an evidenced law can be falsified and fixed.

   • Evidence makes legislation a testable hypothesis.

Why it works: 12 principles

1. Loyalty to data — claims stand or fall on quality, not source.

2. Causal identification — it distinguishes correlation from cause.

3. Effect sizes — it asks not just whether, but how much.

4. Provenance — it attributes every fact to a traceable source.

5. Uncertainty honesty — it reports confidence, not just conclusions.

6. Independence — it owes nothing to the interest that benefits.

7. Falsifiability — it names what would prove the claim wrong.

8. Replication weighting — it trusts findings that reproduce over one-off results.

9. Conflict reconciliation — it resolves contradictory studies rather than cherry-picking one.

10. Robustness balance — it weighs novel findings against established ones.

11. Relevance filtering — it prefers evidence from comparable populations and contexts.

12. Method transparency — it exposes how a conclusion was reached, not just the conclusion.

Three major patterns of how it works

1. Retrieve → weigh → cite

   • Retrieve the relevant research

   • Weigh it by quality and relevance

   • Cite it against the specific claim

2. Synthesise → reconcile → report

   • Synthesise findings across studies

   • Reconcile conflicting results

   • Report a confidence-weighted conclusion

3. Verify → flag → correct

   • Verify each cited source exists and says what is claimed

   • Flag fabrication and overreach

   • Correct before the claim is acted on

Key mechanisms and how they work

A. The standing evidence base

• A vast, mostly unread body of rigorous research already exists.

• Example: The UK’s What Works network spans policy areas accounting for hundreds of billions in public spending, synthesising evidence for decision-makers—proof that the supply of evidence already outstrips the bandwidth to use it.

B. Institutionalised evidence mandates

• Governments have legislated the demand for evidence even where the labour is scarce.

• Example: The bipartisan U.S. Foundations for Evidence-Based Policymaking Act required agencies to build evidence-building plans and appoint Chief Evaluation Officers—an explicit statutory demand for grounding that agents can help supply.

C. Verification against hallucination

• The grounding faculty fails catastrophically if the “evidence” is invented.

• Example: In Mata v. Avianca, lawyers filed a brief citing six entirely fabricated precedents produced by a chatbot; in a separate case, an expert defending a deepfake statute filed sworn testimony with AI-hallucinated citations and was excluded—proof that an evidence agent without a verification layer is worse than none.

How AI changes the game: definition

AI turns evidence from a scarce, lobbyist-supplied subsidy into an abundant, on-demand, client-free resource—retrieving and weighing the research behind any claim in seconds—while introducing a new failure mode: confident fabrication that must be caught before it is cited.

In short: the subsidy finally has no master.

Four principles of how AI changes the game

1. From supplied to retrieved — from facts handed over by an interest to facts pulled from the record.

2. From assertion to citation — from “studies show” to a traceable source.

3. From scarce to continuous — from a one-off literature review to standing grounding.

4. From trust to verification — from believing the output to checking its provenance.

Advantages and disadvantages

Advantages

1. Breaks the lobbyist monopoly on policy information.

2. Grounds every claim in a traceable, weighable source.

3. Reports uncertainty instead of false certainty.

4. Makes law a falsifiable hypothesis that can be corrected.

Disadvantages

1. Hallucinated citations can launder fabrication as scholarship.

2. Automation bias leads officials to over-trust the cited output.

3. Evidence informs but cannot settle value disagreements—it can smuggle values as facts.

4. The training data and the model’s tuner both shape what counts as “evidence.”

6) Simulation

Definition

Simulation is the faculty of testing a law against a model of the world before it becomes binding—war-gaming its effects on a synthetic population and economy to expose the second- and third-order consequences a drafter never imagined.

It functions as the legislature’s staging-environment layer: the missing test harness that lets a society run a law in a sandbox—surfacing the gaming, the perverse incentives, and the distributional losers in silico—before it ships to millions.

Place in lawmaking: 5 aspects

1. The consequence engine

   • Laws fail at the second order, not the first.

   • Simulation reveals the downstream effects.

2. The gaming detector

   • Every rule is an optimisation target for those it binds.

   • Simulation surfaces the evasion in advance.

3. The distributional X-ray

   • Aggregate effects hide who wins and who loses.

   • Simulation shows the losers before the vote.

4. The rollback substitute

   • Law has no easy undo; mistakes are costly.

   • Simulation is the cheap rehearsal that prevents them.

5. The behavioural realism layer

   • People respond, adapt, and evade.

   • Simulation models behaviour, not just arithmetic.

Why it works: 12 principles

1. Foresight — it moves error discovery before enactment.

2. Behavioural modelling — it captures how people actually respond.

3. Distributional resolution — it disaggregates winners and losers.

4. Adversarial testing — it lets the rule be gamed in safety.

5. Cheapness of failure — a failed simulation costs nothing.

6. Scenario range — it explores many futures, not one forecast.

7. Iteration — it lets the law be revised before it bites.

8. Assumption transparency — it states what the model takes for granted.

9. Sensitivity analysis — it shows which assumptions the result depends on.

10. Emergence capture — it surfaces effects no one deliberately designed in.

11. Calibration — it is checked against real-world outcomes where they exist.

12. Comparability — it scores the proposal against the status quo and alternatives on a common scale.

Three major patterns of how it works

1. Model → run → observe

   • Model the population and economy

   • Run the proposed law against it

   • Observe the emergent effects

2. Perturb → adapt → expose

   • Perturb the system with the new rule

   • Let simulated agents adapt and evade

   • Expose the gaming behaviour

3. Score → compare → revise

   • Score outcomes across scenarios

   • Compare against the status quo

   • Revise the law before enactment

Key mechanisms and how they work

A. Microsimulation and rules-as-code

• Tax and benefit law is already tested against synthetic populations before scoring.

• Example: France’s OpenFisca encodes tax-and-benefit law as executable code so a reform can be simulated before it is passed; open successors such as PolicyEngine put the same capability in a citizen’s browser.

B. Agent-based models of the economy

• Whole economies can be simulated as interacting agents for “what-if” policy design.

• Example: Central banks, including the Bank of England, have moved agent-based macroeconomic models from the seminar room into the operating toolkit, and the EU funded agent-based engines explicitly for policy design.

C. Behavioural and synthetic-population simulation

• Generative agents now reproduce real human responses closely enough to poll.

• Example: A Stanford study built generative agents of over a thousand real people from interviews; the agents reproduced their human counterparts’ survey answers about 85% as accurately as the humans reproduced their own answers two weeks later. In a separate model, simulated workers spontaneously learned to avoid a tax code—surfacing the gaming before it could hit the real economy.

How AI changes the game: definition

AI turns simulation from siloed, expert-only microsimulation into a general staging environment for any law—modelling behaviour, gaming, and distribution across a synthetic society—while risking over-trust in models that are fragile, gameable, and only as honest as their assumptions.

In short: law finally gets a test harness.

Four principles of how AI changes the game

1. From arithmetic to behaviour — from static scoring to modelled human response.

2. From aggregate to distributional — from a single number to who-wins-who-loses.

3. From forecast to war-game — from one projection to adversarial scenarios.

4. From narrow domains to all law — from tax-only microsimulation to general policy testing.

Advantages and disadvantages

Advantages

1. Moves the discovery of unintended consequences before enactment.

2. Surfaces gaming and evasion in safety.

3. Reveals distributional losers the aggregate hides.

4. Makes failure cheap and revision routine.

Disadvantages

1. Models are fragile; a buggy simulation can mislead with authority.

2. Any simulated metric becomes a target interests will reverse-engineer and game.

3. Synthetic populations inherit the biases of their training data.

4. False confidence in a model can be more dangerous than honest uncertainty.

7) Composition

Definition

Composition is the faculty of converting settled intent into precise, conflict-free statutory text—translating a policy decision into legal language that says exactly what it means and collides with nothing it should not.

It functions as the legislature’s drafting layer: the final translation from what we decided to what the statute says, where loopholes, ambiguities, and litigation are either prevented or created.

Place in lawmaking: 5 aspects

1. The intent translator

   • A decision is not yet a law until it is text.

   • Composition renders intent into enforceable language.

2. The loophole closer

   • Imprecise drafting is where evasion lives.

   • Composition tightens the text against exploitation.

3. The consistency keeper

   • New text must cohere with the existing corpus.

   • Composition harmonises language across statutes.

4. The accessibility shaper

   • Law that no citizen can read loses legitimacy.

   • Composition can render text in plain language too.

5. The throughput multiplier

   • Drafting capacity caps how much law a body can produce.

   • Composition raises that ceiling—for better or worse.

Why it works: 12 principles

1. Precision — it says exactly what is meant.

2. Consistency — it aligns with definitions already in force.

3. Completeness — it anticipates the cases the rule must cover.

4. Conflict-freedom — it avoids collision with existing law.

5. Traceability — it links each clause to its intent.

6. Revisability — it red-lines and iterates quickly.

7. Legibility — it can produce a human-readable companion.

8. Speed — it produces a working draft in minutes, not weeks.

9. Edge-case coverage — it anticipates the situations a rule must handle.

10. Definitional discipline — it reuses terms already defined in the corpus.

11. Enforceability — it writes text that can actually be applied and adjudicated.

12. Plain-language parity — it keeps the readable version faithful to the legal one.

Three major patterns of how it works

1. Intent → draft → red-line

   • Capture the settled intent

   • Draft the statutory language

   • Red-line against corpus and edge cases

2. Generate → check → harmonise

   • Generate candidate text

   • Check for conflicts and loopholes

   • Harmonise with existing definitions

3. Translate → simplify → publish

   • Translate legal text into plain language

   • Simplify for public comprehension

   • Publish both versions together

Key mechanisms and how they work

A. Direct AI drafting by elected officials

• Legislators are already drafting real bills with language models.

• Example: A Porto Alegre councillor had a chatbot draft a municipal ordinance from a forty-nine-word prompt and the council passed it unanimously; a Massachusetts state senator used the same tools to draft an AI-regulation bill that he said got him “about seventy percent of the way there.”

B. Corpus-grounded drafting assistants

• National drafting offices are building assistants trained on the full body of law.

• Example: The UK’s Office of the Parliamentary Counsel built a drafting assistant grounded in roughly 1.5 million sections of legislation and tens of thousands of court cases, generating explanatory material and supporting precise legal language.

C. The volume warning

• Cheap drafting raises throughput, which is not the same as raising quality.

• Example: Lowering the cost of writing bills has already produced a flood—on the order of a thousand AI-related bills introduced in a few months of a single U.S. session—demonstrating that composition without judgment yields volume, not law.

How AI changes the game: definition

AI turns composition from a scarce, specialist bottleneck into an abundant, on-demand capability—drafting and red-lining precise statutory text grounded in the corpus—while collapsing the cost of producing bad law just as fast as good law.

In short: drafting stops being the bottleneck—and judgment becomes it.

Four principles of how AI changes the game

1. From scarce to abundant — from a specialist queue to on-demand drafting.

2. From blank page to grounded draft — from starting cold to starting from the corpus.

3. From opaque to legible — from impenetrable text to a plain-language companion.

4. From production-limited to judgment-limited — the constraint moves from writing to deciding.

Advantages and disadvantages

Advantages

1. Collapses the cost and delay of precise drafting.

2. Closes loopholes by red-lining against the whole corpus.

3. Produces plain-language versions that raise legitimacy.

4. Gives a back-bencher the drafting capacity of a leadership office.

Disadvantages

1. Cheap drafting floods the system with volume over quality.

2. Whoever owns the drafting model can steer statutory language at scale.

3. Undisclosed AI authorship raises real accountability and legitimacy questions.

4. Fluent text can mask substantive errors a human would have caught.

8) Constituent Sensing

Definition

Constituent Sensing is the faculty of hearing what the public actually needs—mapping the real preferences, burdens, and priorities of citizens at scale, and separating genuine signal from manufactured noise.

It functions as the legislature’s input layer: the mechanism that lets a representative perceive the people they serve directly, rather than through the filter of whoever can afford to manufacture the loudest voice.

Place in lawmaking: 5 aspects

1. The representation anchor

   • A representative who cannot hear the represented is one in name only.

   • Sensing restores the direct line between citizen and lawmaker.

2. The signal–noise filter

   • Organised campaigns drown out individual citizens.

   • Sensing separates substance from orchestrated volume.

3. The burden detector

   • Citizens pay a silent “time tax” they rarely write letters about.

   • Sensing surfaces friction, not only stated opinion.

4. The preference map

   • Opinion is distributed unevenly across issues and groups.

   • Sensing renders what the public actually wants, by segment.

5. The astroturf shield

   • Manufactured, bot-driven, and duplicated input corrupts the record.

   • Sensing detects and discounts it.

Why it works: 12 principles

1. Directness — it perceives citizens without an intermediary.

2. Scale — it processes millions of inputs, not a sampled few.

3. Signal extraction — it separates substance from form-letter volume.

4. Authenticity detection — it flags fabricated or duplicated comments.

5. Disaggregation — it sees subgroups, not just the average.

6. Inclusivity — it hears those who lack organised representation.

7. Multilingual reach — it understands input in any language.

8. Continuity — it listens between elections, not only at them.

9. Burden sensitivity — it detects friction and “time tax,” not only opinion.

10. Proportionality — it weights by genuine prevalence, not manufactured volume.

11. Privacy preservation — it can aggregate without exposing individuals.

12. Responsiveness — it routes real concerns to the relevant decision.

Three major patterns of how it works

1. Collect → dedupe → classify

   • Collect inputs across every channel

   • Remove duplicates and astroturf

   • Classify by topic, sentiment, and segment

2. Aggregate → weight → surface

   • Aggregate by genuine prevalence

   • Weight by authenticity

   • Surface the real distribution of need

3. Detect → verify → route

   • Detect a genuine concern

   • Verify it is organic

   • Route it to the relevant faculty

Key mechanisms and how they work

A. Public-comment analysis at scale

• Agencies receive millions of comments; AI categorises, deduplicates, and flags bot-generated versus substantive input.

• Example: Federal comment-analysis pipelines now compress what once took weeks of manual review into hours, triaging millions of public comments on proposed rules into topics and genuine-versus-duplicate buckets.

B. Astroturf and fraud detection

• Natural-language analysis reveals manufactured campaigns hiding inside the record.

• Example: Of the roughly 22 million comments on the FCC’s net-neutrality repeal, analysis found about 18 million were fake, with fewer than 800,000 genuinely organic—exactly the noise a sensing agent must strip out.

C. Direct citizen-engagement channels

• Standing platforms let citizens shape decisions between elections, not only at them.

• Example: Singapore’s REACH e-engagement platform gathers citizen feedback on policies directly—a model for structured, ongoing constituent input rather than episodic polling.

How AI changes the game: definition

AI turns constituent sensing from a sampled, gameable trickle into scaled, authenticated, real-time perception of the public—hearing millions directly and filtering the manufactured—while raising the danger of ever more convincing synthetic “citizens.”

In short: the representative can finally hear the represented.

Four principles of how AI changes the game

1. From sample to population — from a few loud voices to the whole distribution.

2. From form-letter to substance — from counting volume to extracting signal.

3. From episodic to continuous — from election-day to always-on listening.

4. From gameable to authenticated — from astroturf-vulnerable to fraud-aware.

Advantages and disadvantages

Advantages

1. Restores the representative’s direct line to the represented.

2. Surfaces silent burdens the vocal never raise.

3. Filters manufactured campaigns from genuine concern.

4. Hears the unorganised, multilingual, and marginalised.

Disadvantages

1. Generative AI also makes synthetic “constituents” cheaper and more convincing.

2. Aggregating citizen input at scale raises surveillance and privacy risks.

3. Sensing can be mistaken for a mandate, bypassing deliberation.

4. What the model labels “noise” may include real but unconventional voices.

9) Deliberation

Definition

Deliberation is the faculty of stress-testing a decision through argument—steelmanning the opposition, surfacing the trade-offs, applying the role-reversal test, and forcing a proposal to survive its strongest objections before it becomes law.

It functions as the legislature’s reasoning layer: the adversarial discipline that converts evidence and simulation into a justified choice, ensuring a law is defended on public reasons rather than passed on tribal reflex.

Place in lawmaking: 5 aspects

1. The objection generator

   • Drafters see the case for, rarely the strongest case against.

   • Deliberation manufactures the best opposing argument.

2. The trade-off namer

   • Every law has costs its sponsors prefer to leave implicit.

   • Deliberation makes who-benefits-and-who-pays explicit.

3. The role-reversal test

   • A rule acceptable from power may be intolerable from opposition.

   • Deliberation tests it from every position.

4. The public-reason filter

   • Justifications by tribe or creed do not bind a plural society.

   • Deliberation demands reasons any citizen could accept.

5. The blind-spot finder

   • Authors cannot see what they did not think of.

   • Deliberation surfaces the unconsidered.

Why it works: 12 principles

1. Adversarial rigour — it attacks the proposal to find its weaknesses.

2. Steelmanning — it builds the strongest version of the opposing case.

3. Impartiality — it judges arguments by merit, not source.

4. Trade-off candour — it names costs, not only benefits.

5. Reversibility test — it checks the rule from every stakeholder’s position.

6. Public reason — it requires justifications independent of tribe or creed.

7. Assumption surfacing — it exposes hidden premises.

8. Perspective breadth — it represents absent and minority viewpoints.

9. Consistency — it treats like cases alike across time and party.

10. Falsification framing — it asks what would change the conclusion.

11. Proportionality — it favours the least-restrictive effective means.

12. Humility — it admits the limits of what is known.

Three major patterns of how it works

1. Propose → attack → defend

   • State the proposal

   • Generate the strongest objections

   • Force a defence or a revision

2. Reframe → reverse → test

   • Reframe from each stakeholder’s view

   • Apply the role-reversal

   • Test for acceptability across positions

3. Expose → weigh → justify

   • Expose the trade-offs

   • Weigh them openly

   • Justify the choice on public reasons

Key mechanisms and how they work

A. Multi-perspective argument generation

• Agents can argue every side of a question, attacking and defending in turn.

• Example: Work on adversarial and multi-agent reasoning shows that pitting models against each other to attack and defend a claim surfaces weaknesses a single pass misses—a standing red-team for legislation.

B. Structured value frameworks

• Explicit principles convert open argument into disciplined judgment.

• Example: The Apolitical Politics framework already codifies the role-reversal test, public reasons, and trade-off candour—the exact criteria a deliberation agent can apply, clause by clause, to every bill.

C. Distributional and objection mapping

• Surfacing who loses, and why, turns abstract objection into specific accountability.

• Example: The same microsimulation and synthetic-population tools that reveal distributional losers feed deliberation by naming, concretely, whose interests a law moves—so the objection is grounded, not rhetorical.

How AI changes the game: definition

AI turns deliberation from a scarce, often-skipped luxury into a standing adversarial process—generating the strongest objections, the role-reversal, and the trade-off map for every proposal—while risking persuasive argument detached from truth.

In short: every bill can be cross-examined before it is passed.

Four principles of how AI changes the game

1. From advocacy to adversary — from arguing one side to attacking every side.

2. From implicit to explicit trade-offs — from hidden costs to a named distribution.

3. From tribal to public reasons — from “our side” to justifications all can assess.

4. From skipped to standing — from rushed votes to routine cross-examination.

Advantages and disadvantages

Advantages

1. Forces a proposal to survive its strongest objections.

2. Makes trade-offs and losers explicit before the vote.

3. Represents absent and minority perspectives.

4. Anchors the decision in public reasons, not reflex.

Disadvantages

1. Fluent argument can persuade without being true—rhetoric outruns evidence.

2. Endless deliberation can become a tactic for delay.

3. The model’s framing of “the other side” embeds its own biases.

4. Manufactured objections can obstruct as easily as improve.

10) Oversight

Definition

Oversight is the faculty of learning whether a law actually worked—monitoring its real-world effects after passage, evaluating it against its stated goals, and triggering revision or repeal when it fails.

It functions as the legislature’s feedback layer: the loop that converts a law from a one-time act into a testable hypothesis, closing the cycle so that legislation learns instead of accumulating as dead, unexamined sediment.

Place in lawmaking: 5 aspects

1. The hypothesis closer

   • A law is a prediction that an intervention will help.

   • Oversight tests whether the prediction held.

2. The ghost-law detector

   • Statutes pass and are never revisited.

   • Oversight finds the dead letters still on the books.

3. The enforcement monitor

   • Text on the page is not the same as a rule applied.

   • Oversight checks whether the law actually bites.

4. The sunset trigger

   • Some laws should expire or be revised on schedule.

   • Oversight flags when their time has come.

5. The learning capture

   • Each law is a lesson for the next.

   • Oversight turns outcomes into institutional memory.

Why it works: 12 principles

1. Falsifiability — it treats every law as a hypothesis with a stated test.

2. Goal anchoring — it measures against the law’s own declared aims.

3. Continuity — it watches outcomes long after the vote.

4. Honesty about misses — it surfaces failure rather than hiding it.

5. Counterfactual measurement — it asks what would have happened otherwise.

6. Enforcement realism — it checks application, not just text.

7. Timeliness — it flags failure early, not after decades.

8. Reversibility — it makes repeal and revision routine.

9. Comparability — it scores outcomes against the original forecast.

10. Independence — it evaluates free of the author’s stake.

11. Transparency — it publishes results, including the failures.

12. Cumulativeness — it feeds lessons forward into future legislation.

Three major patterns of how it works

1. Measure → compare → judge

   • Measure real outcomes

   • Compare to stated goals

   • Judge success or failure

2. Monitor → flag → trigger

   • Monitor enforcement and effect

   • Flag drift or failure

   • Trigger review

3. Evaluate → publish → feed-forward

   • Evaluate against the forecast

   • Publish the result

   • Feed lessons into the next law

Key mechanisms and how they work

A. Ex-post evaluation at scale

• Most laws are never rigorously revisited after passage.

• Example: Scoping reviews of ex-post legislative evaluation find the societal impact of laws is “rarely measured” after enactment, leaving “ghost laws” on the books—precisely the gap a standing oversight agent closes.

B. Institutional evaluation mandates

• Governments have legislated the demand for post-hoc evidence even where the labour is scarce.

• Example: The U.S. Foundations for Evidence-Based Policymaking Act requires agencies to appoint Chief Evaluation Officers and assess the coverage and quality of their evaluations—an oversight mandate agents can help fulfil continuously.

C. Corpus pruning as oversight

• AI can identify obsolete or redundant law for repeal.

• Example: Stanford’s RegLab statutory-research system and Ohio’s code review both used AI to find outdated statutes—oversight applied to the existing corpus, surfacing what should be revised or removed.

How AI changes the game: definition

AI turns oversight from a rare, after-the-fact audit into continuous, automated evaluation—measuring every law against its goals and flagging failure early—while risking metric-driven judgment that mistakes the measurable for the meaningful.

In short: law finally learns from its own results.

Four principles of how AI changes the game

1. From one-time act to standing hypothesis — from “passed” to “still working?”

2. From decades to real-time — from belated review to early warning.

3. From hidden to published — from buried failure to transparent result.

4. From isolated to cumulative — from each law alone to lessons that compound.

Advantages and disadvantages

Advantages

1. Closes the loop so legislation learns from outcomes.

2. Surfaces ghost laws and obsolete statutes for repeal.

3. Makes failure visible early, when it is still cheap to fix.

4. Feeds concrete lessons into the next round of lawmaking.

Disadvantages

1. Metric-driven oversight can optimise the measurable and miss the meaningful.

2. Continuous monitoring carries surveillance and privacy risks.

3. Evaluation framed by the model can embed its own definition of “success.”

4. Automated repeal flags could strip protective laws under the banner of efficiency.

Action plan

The Augmented Legislator is not built by buying a chatbot. It is built by installing the Legislative Intelligence Stack as public infrastructure, owned by the institution and the citizen rather than by a vendor or an interest. The sequence matters: comprehension and evidence first, simulation and composition last, with legitimacy designed in at every layer.

Phase 1: Give the legislature its own eyes and ears (Comprehension + Evidence + Constituent Sensing)

1. Deploy a public statutory-comprehension model. Stand up a corpus-grounded system over the full body of law, with conflict detection, dead-letter flagging, and provenance on every clause—so no representative ever again votes on text no one has read.

2. Install a client-free evidence layer. Pair every bill with an evidence agent that retrieves, weighs, and cites the research behind each claim—with a mandatory verification step that catches fabricated citations before they reach the floor.

3. Stand up a constituent-sensing channel. Collect, deduplicate, and authenticate public input at scale, filtering astroturf, so the representative hears the real distribution of need—not the loudest manufactured campaign.

4. Mandate provenance and uncertainty. Require that every machine-supplied fact carry its source and confidence, and every machine-flagged conflict carry a plain-language explanation.

Phase 2: Make prioritisation deliberate (Significance + Tractability)

5. Build a standing problem-triage register. Score candidate problems by reach, severity, urgency, and reversibility, publish the ranking, and force any deviation from it to be justified.

6. Attach a tractability estimate to every proposal. Require a modelled effect size and an implementation-friction appraisal—grounded in the global trial record—before a bill consumes a hearing.

7. Log the dropped problems. Publish what the triage de-prioritised and why, so significance failures are visible, not silent.

Phase 3: Test before you ship (Diffusion + Simulation)

8. Stand up a diffusion service. Maintain a live map of analogous policies across jurisdictions, with measured outcomes and documented failures, so every draft starts from what already worked.

9. Require a staging run for material laws. Before enactment, war-game the bill against a microsimulation, an economic model, and a synthetic population—surfacing gaming, perverse incentives, and distributional losers—and publish the result.

10. Adopt rules-as-code. Encode the operative provisions as executable logic so the law can be simulated, queried, and re-tested as conditions change.

Phase 4: Deliberate, then draft with judgment (Deliberation + Composition + Legitimacy)

11. Cross-examine every bill. Before drafting, require a deliberation pass that generates the strongest objections, names the trade-offs and the losers, and applies the role-reversal test—so the decision rests on public reasons, not on whoever held the floor.

12. Give every representative their own drafting agent. Open-weighted where possible, auditable, logged, and grounded in the corpus—so the drafting subsidy that lobbyists once monopolised belongs to the back-bencher and the public, not the best-funded interest.

13. Bind the Stack to the Apolitical Politics oath. The agent supplies the mechanism, the evidence, the simulation, and the text; the human owns the role-reversal test, the public reasons, the candid naming of who benefits and who pays, the preference for reversible choices—and the vote. The machine does the is; the accountable human, within inviolable rights, chooses the ought.

14. Distribute ownership as the antidote to capture. Refuse a single official model. The defence against an agent that writes laws for its owner is many competing, auditable agents owned by representatives, parties, and citizens—legitimacy through plurality, not monopoly.

Phase 5: Oversee, measure, and learn (Oversight)

15. Treat every law as a standing hypothesis. Attach stated goals and a falsification test at passage; monitor real-world outcomes and enforcement; flag ghost laws and trigger the revision or repeal of what fails.

16. Publish the guarantees. Track time-to-evidence before a vote, the share of bills with a published simulation and an ex-post evaluation plan, the proportion of proven cross-jurisdictional policy actually surfaced, and the falsification record of laws once passed.

17. Keep a human override at every layer. When a model fails, revert to human procedure; treat every automated step as advisory to an accountable person.

Named deliverable: the Legislative Intelligence Stack Charter—a per-office specification of the ten capabilities, their ownership and audit rules, their binding to the legislator’s oath, and the published outcome metrics by which the citizen judges whether the augmented legislature is, in fact, governing better. The law was never too complex for democracy. Democracy was simply never given enough mind. The Stack gives it one—without ever taking away the vote.

What it has not yet built is the layer that turns those foundations into a state that serves the citizen unbidden. This is that architecture, division by division—what exists right now, where it is slow and fragmented, and what each layer becomes when agents are first-class workers.

For twenty years the Czech Republic built the foundations of a digital state and then stopped one layer short. The country has the base registers—ROB for people, ROS for organizations, RÚIAN for addresses, RPP for the rights and obligations of every agenda. It has a national identity in NIA and the widely used BankID, and since 2024 mobile documents in eDoklady. It has data mailboxes in the ISDS, Czech POINT counters, the Portál občana, and a legally enabled connected data fund, the propojený datový fond, with its reference interface ISZR and eGSB/ISSS. Since 2023 it even has a single owner of the digital agenda, the Digitální a informační agentura (DIA), gestor of 21 agendas and active in 54 more, operator of more than 41 public information systems. The foundations are real, and most countries envy them. The missing piece is not another register or another portal. It is the orchestration layer that makes the state assemble itself around the citizen—and the citizen stop being the integrator of the state. This article describes that architecture in twelve divisions.

The first division is Identity. Today a Czech citizen proves who they are through NIA, BankID, the Mobile Key, or eDoklady—a real, working identity layer, if a fragmented one. The agentic state extends it to the European identity wallet and, crucially, to the agents themselves, so the state can tell its own sovereign agent from an impostor and a citizen’s agent from a stranger’s.

The second division is The Registers. ROB, ROS, RÚIAN, and RPP are the country’s authoritative sources of truth—a genuine asset. But the Informační koncepce itself admits there is no unified data model and no data stewards for all entities. The agentic state makes the registers the single, governed source every agent queries—and never copies.

The third division is The Data Fund. The propojený datový fond, with ISZR and eGSB/ISSS, is the plumbing that lets agencies query one another instead of demanding the same fact twice. It exists in law and in part in practice. The agentic state turns it into the real-time query fabric on which “ask once, never copy” becomes architecture, not aspiration.

The fourth division is The Agenda Model. RPP—the register of rights and obligations—already encodes every agenda and service of the Czech state in machine-readable form. It is the country’s most underused asset. The agentic state makes RPP the law the agent reasons over, the formal model of what the state may and must do for a citizen in any situation.

The fifth division is The Channels. Datové schránky, Czech POINT, the Portál občana, and eDoklady are the front doors of the digital state—but they are doors the citizen must find, open, and walk through in the right order. The agentic state collapses them into one continuous conversation on any surface: voice, text, app, counter.

The sixth division is The Orchestration Layer. This is the piece that does not exist yet, and the reason the others underperform. There is no layer that composes agents dynamically across the 75 agendas the DIA touches to serve a single life event. Building it is the central act of the agentic state—the difference between a digital state and an agentic one.

The seventh division is The Runtime. The Czech state runs no production AI agents at scale today; it has a National AI Strategy 2030, a draft implementing law, and the EU AI Act to transpose. The agentic state gives it a sovereign, EU-hostable, inspectable model runtime, AI-Act-conformant by construction—cognition the state owns and can audit.

The eighth division is The Approval Layer. The správní řád requires that a human official decide matters of a citizen’s rights. This is not an obstacle—it is the architecture’s keystone. The agentic state keeps the official holding the pen: the agent prepares, the official approves, and the system runs inside existing law.

The ninth division is Decision and Audit. Today an administrative decision produces a file; an agentic decision must produce a reason and a record. This division builds the reason traces, decision logs, and appeal paths that make every agentic decision contestable—the SCHUFA principle rendered as infrastructure.

The tenth division is Governance and Mandate. The DIA, the Digitální Česko program, and Act 12/2020 on the right to digital services already define who owns the digital agenda. The agentic state adds the founding act—a government resolution—and the expert center to execute it, turning a fragmented mandate into a delivery engine.

The eleventh division is Infrastructure. The eGovernment cloud and the CLOUDIA private cloud are where Czech public IT already runs. The agentic state turns this into the sovereign compute and hosting floor on which the model runtime stands—because a state that cannot host its own cognition does not own it.

The twelfth division is Resilience. The current state is fragmented but, precisely because it is fragmented, it fails locally. The agentic state must not trade that accidental resilience for a brittle monoculture. This division guarantees the manual fallback, model diversity, data stewardship, and data quality that keep the whole architecture trustworthy.

This article is a field guide to the architecture of the Czech Agentic State. It describes twelve divisions, and dissects each one identically—the Layer it defines, its Current state in the Czech Republic right now, the Gap that holds it back, the seven properties of the agentic target, three patterns of how it works, ten components, the four “from → to” shifts it makes, the concrete moves to build it, and an honest ledger of advantages and risks. It closes with a phased Action Plan anchored to the DIA and the existing Czech stack, and a named deliverable: the Czech Agentic State Architecture Charter.

Summary

01 :: Identity

• The layer: who the citizen—and the agent—is.

• What exists now: NIA, BankID, Mobile Key, eDoklady (2024).

• The gap: fragmented identity means; no identity for agents.

• The agentic target: the EU wallet plus verifiable identity for agents.

02 :: The Registers

• The layer: the authoritative sources of truth.

• What exists now: ROB, ROS, RÚIAN, RPP under the DIA.

• The gap: no unified data model, no data stewards for all entities.

• The agentic target: one governed source every agent queries, never copies.

03 :: The Data Fund

• The layer: the exchange fabric between agencies.

• What exists now: the propojený datový fond, ISZR, eGSB/ISSS.

• The gap: partial implementation; uneven real-time query.

• The agentic target: the live fabric that makes “ask once, never copy” real.

04 :: The Agenda Model

• The layer: the machine-readable rights and obligations of the state.

• What exists now: RPP encodes every agenda and service.

• The gap: RPP is underused as a reasoning substrate.

• The agentic target: the formal law the agent reasons over.

05 :: The Channels

• The layer: the citizen’s front door to the state.

• What exists now: datové schránky, Czech POINT, Portál občana, eDoklady.

• The gap: many doors the citizen must find and sequence.

• The agentic target: one continuous conversation on any surface.

06 :: The Orchestration Layer

• The layer: dynamic composition of agents across agendas.

• What exists now: nothing—this is the missing piece.

• The gap: no cross-agenda orchestration exists today.

• The agentic target: the layer that assembles the state around a life event.

07 :: The Runtime

• The layer: the model cognition the state runs on.

• What exists now: strategy and draft law; no production agents at scale.

• The gap: no sovereign, inspectable runtime.

• The agentic target: EU-hostable, AI-Act-conformant cognition the state owns.

08 :: The Approval Layer

• The layer: the human decision on a citizen’s rights.

• What exists now: the správní řád requires an official to decide.

• The gap: no agent-to-official preparation workflow.

• The agentic target: the official holds the pen; the agent prepares.

09 :: Decision and Audit

• The layer: reasons, records, and recourse.

• What exists now: administrative files, limited machine reasons.

• The gap: no reason traces or contestability for automated steps.

• The agentic target: every decision carries a reason and an appeal.

10 :: Governance and Mandate

• The layer: who owns and authorizes the transformation.

• What exists now: DIA, Digitální Česko, Act 12/2020.

• The gap: fragmented delivery; lagged statutory deadlines.

• The agentic target: a founding resolution and an expert delivery center.

11 :: Infrastructure

• The layer: the compute and hosting floor.

• What exists now: eGovernment cloud, CLOUDIA private cloud.

• The gap: not yet a sovereign runtime host.

• The agentic target: the sovereign compute floor under the cognition.

12 :: Resilience

• The layer: the guarantees that keep the architecture trustworthy.

• What exists now: accidental resilience from fragmentation.

• The gap: weak data quality, no manual-fallback guarantee for agents.

• The agentic target: manual fallback, diversity, stewardship, quality.

The Twelve Divisions

01 :: Identity

The Layer

The Identity division establishes, beyond doubt, who the citizen is and—newly—who an agent is, so that every action in the agentic state is bound to an authenticated person and an authenticated agent acting on their behalf.

It functions as the root of trust of the whole architecture: nothing above it can be trusted further than the identity beneath it.

Current state :: what exists now in the Czech Republic

1. NIA (Národní bod pro identifikaci a autentizaci)—the national identity broker, operated by the DIA.

2. BankID—bank-issued identity, the most widely used real-world means of proving who you are.

3. The Mobile Key (Mobilní klíč eGovernmentu) and eObčanka—state-issued electronic identity.

4. eDoklady (2024)—mobile identity and documents, the newest and fastest-growing channel.

5. eIDAS cross-border recognition, with the EU Digital Identity Wallet on the horizon.

The gap :: where we are slow or fragmented

• Identity is fragmented across several means with uneven coverage and user experience.

• There is no identity for agents at all—the state cannot yet authenticate a software actor acting for a citizen or for itself.

• Authentication failure has no guaranteed graceful fallback designed for an agentic flow.

The agentic target :: 7 properties of the layer done right

1. Unified — one coherent identity experience across NIA, BankID, and eDoklady.

2. Wallet-based — built on the EU Digital Identity Wallet (eIDAS2) as the default.

3. Agent-aware — verifiable credentials authenticate agents, not only humans.

4. Delegable — a citizen can authorize an agent to act, revocably and provably.

5. Selective — minimal disclosure; the agent learns only the attribute it needs.

6. Inclusive — a guaranteed non-digital path for those who cannot authenticate.

7. Auditable — every authentication and delegation is logged and inspectable.

Three patterns of how it works

1. Authenticate → delegate → act

   • The citizen authenticates

   • They delegate authority to an agent, provably

   • The agent acts within that mandate

2. Human identity → agent identity → bound action

   • The person is identified

   • The agent is identified

   • Each action is bound to both

3. Request → minimal disclosure → log

   • An attribute is requested

   • The minimum is disclosed

   • The disclosure is logged

Ten components :: the building blocks

1. NIA (the national identity broker)

2. BankID (bank-issued identity)

3. Mobile Key / eObčanka (state electronic identity)

4. eDoklady (mobile documents)

5. The EU Digital Identity Wallet (eIDAS2)

6. Verifiable credentials for agents (DIDs/VCs)

7. A delegation registry (who authorized which agent)

8. Selective-disclosure protocols (minimal attributes)

9. A guaranteed non-digital fallback (inclusion)

10. Authentication and delegation logs (audit)

The shift :: four “from → to” moves

1. From fragmented means to a unified wallet

   • One coherent identity built on eIDAS2.

2. From human-only to human-and-agent identity

   • Agents are authenticated actors too.

3. From implicit to provable delegation

   • A citizen’s authorization of an agent is explicit and revocable.

4. From all-or-nothing to selective disclosure

   • The agent learns the minimum, nothing more.

How to build it

A. Make the EU wallet the spine, not a parallel track

• Converge NIA, BankID, and eDoklady onto the EU Digital Identity Wallet rather than adding a fifth silo.

• Example: issue agent-delegation credentials through the same wallet a citizen already uses for eDoklady.

B. Give agents verifiable identity

• Issue every sovereign and citizen agent a credential the state can check.

• Example: a state benefits-agent carries a credential that an agency can verify before accepting its request.

C. Guarantee the fallback in design

• Ensure no right depends on a successful authentication; a human path always exists.

• Example: a Czech POINT counter remains a full, non-digital route to every agentic service.

Advantages and risks

Advantages

1. A single, trustworthy root for the whole architecture.

2. Agents that can be authenticated and held to a mandate.

3. Provable, revocable delegation from citizen to agent.

4. Minimal disclosure as a built-in privacy property.

Risks

1. Converging existing means is politically and technically hard.

2. Agent identity is an immature standard with real attack surface.

3. Delegation, if abused, lets an agent overreach a citizen’s intent.

4. Wallet dependence concentrates risk in one credential store.

What needs to be done :: implementation backlog

The work to implement (current → future):

1. Converge NIA, BankID, the Mobile Key, and eDoklady behind one identity experience aligned to the EU Digital Identity Wallet (eIDAS2)—stop adding silos.

2. Issue the EU Digital Identity Wallet to citizens and make it the default credential for agentic flows.

3. Build an agent-identity scheme—verifiable credentials (DIDs/VCs) for every state agent and every citizen agent.

4. Stand up a delegation registry recording which agent a citizen authorized, with provable, revocable mandates.

5. Implement selective disclosure (minimal-attribute requests) across all identity means.

6. Define and deploy a guaranteed non-digital fallback (the Czech POINT route) so authentication failure never denies a right.

7. Centralize authentication and delegation logging, citizen-visible.

8. Certify the layer against eIDAS2 and the AI Act wherever identity feeds a decision.

Principles to apply:

• Inclusive Legibility — everyone recognized; the fallback is a right.

• Sovereign-European Runtime — the EU wallet, not a foreign credential broker.

• Minimization Is the Privacy Firewall — selective disclosure by default.

• The Citizen Is No Longer the Integrator — delegation is what lets an agent act for a citizen at all.

The architecture state change:

• Today: four-plus separate identity means, no identity for agents, no provable delegation. → Future: one wallet-based identity, verifiable agent credentials, provable and revocable delegation, minimal disclosure, a guaranteed human fallback.

The advantages of getting there:

1. A single, trustworthy root of trust for the entire architecture.

2. Agents can lawfully and revocably act for citizens—unlocking every higher division.

3. Privacy improves (minimal disclosure) even as capability grows.

4. eIDAS2 alignment delivers cross-border, EU-interoperable identity.

5. No citizen is excluded—the fallback is guaranteed in design.

02 :: The Registers

The Layer

The Registers division holds the authoritative facts of the state—who exists, which organizations exist, where places are, and what every agenda may and must do—so that the whole agentic state reasons from one governed source of truth rather than a thousand drifting copies.

It functions as the source of truth of the architecture: the ground on which every agent’s decision stands.

Current state :: what exists now in the Czech Republic

1. ROB (Registr obyvatel)—the authoritative register of residents.

2. ROS (Registr osob)—the authoritative register of organizations and businesses.

3. RÚIAN—the register of territorial identification, addresses, and real estate.

4. RPP (Registr práv a povinností)—the register of agendas, rights, and obligations.

5. The base registers are operated by the DIA through the reference interface (ISZR, eGSB/ISSS).

The gap :: where we are slow or fragmented

• The Informační koncepce admits there is no unified data model and no systematic categorization of data across agendas.

• Data stewards are not established for all data entities, and there is no systematic data-quality monitoring.

• Many agendas still do not keep data as the law requires, weakening the authority of the source.

The agentic target :: 7 properties of the layer done right

1. Authoritative — one designated, legally binding source per class of fact.

2. Modeled — a unified data model spanning agendas, not per-system silos.

3. Stewarded — a named data steward accountable for every entity.

4. Quality-monitored — systematic, continuous data-quality measurement.

5. Queryable — readable in real time through the reference interface.

6. Governed — purpose-bound, logged access to every fact.

7. Never copied — agents read the source; they do not shadow it.

Three patterns of how it works

1. Designate → steward → maintain

   • One register is authoritative

   • A steward owns its quality

   • It is maintained as the source

2. Query → use → forget

   • An agent queries the register

   • Uses the fact

   • Retains no copy

3. Model → categorize → govern

   • Data is modeled uniformly

   • Categorized by agenda

   • Access is governed

4. (patterns held to three)

Ten components :: the building blocks

1. ROB (residents)

2. ROS (organizations)

3. RÚIAN (addresses and real estate)

4. RPP (agendas, rights, obligations)

5. The reference interface (ISZR, eGSB/ISSS)

6. A unified data model (the missing piece)

7. Designated data stewards (accountable owners)

8. Data-quality monitoring (systematic, continuous)

9. Purpose-binding and access logs (governance)

10. A no-copy policy (query, never shadow)

The shift :: four “from → to” moves

1. From siloed data to a unified model

   • One model spans agendas instead of per-system fragments.

2. From unowned data to stewarded data

   • Every entity has an accountable steward.

3. From unmeasured to quality-monitored

   • Data quality is measured systematically.

4. From copies to queries

   • Agents read the authoritative source and never duplicate it.

How to build it

A. Appoint data stewards before building agents

• Make the registers trustworthy by naming accountable owners for every entity.

• Example: the Informační koncepce’s own diagnosis—missing data stewards—becomes the first thing the agentic program fixes.

B. Build the unified data model

• Replace per-system data definitions with one model spanning agendas.

• Example: a single, governed model that ROB, ROS, RÚIAN, and RPP share, so an agent reasons consistently across them.

C. Forbid the shadow copy

• Mandate that agents query the registers and never maintain duplicates.

• Example: a benefits-agent reads the resident’s address from ROB at decision time rather than caching it.

Advantages and risks

Advantages

1. One consistent truth for the entire state.

2. Accountable stewardship and measurable quality.

3. Minimization—no shadow copies to leak.

4. A reliable substrate for cross-agenda composition.

Risks

1. A wrong fact in a register propagates everywhere instantly.

2. Establishing stewardship across agencies is politically hard.

3. Real-time query availability becomes mission-critical.

4. A unified model is a large, slow undertaking that must not stall delivery.

What needs to be done :: implementation backlog

The work to implement (current → future):

1. Appoint named data stewards for every data entity across ROB, ROS, RÚIAN, and RPP—close the Informační koncepce’s own diagnosed gap.

2. Build the unified data model spanning agendas, replacing per-system definitions.

3. Stand up systematic, continuous data-quality monitoring with published metrics.

4. Bring every agenda’s data into legal compliance with record-keeping requirements.

5. Designate one legally authoritative source per class of fact.

6. Enforce a no-copy policy—agents query the registers, never shadow them.

7. Add purpose-binding and access logging at the register level.

8. Build identity resolution linking a citizen reliably to the authoritative record.

Principles to apply:

• The Register Is the Single Source of Truth — one governed source, not many copies.

• Ask Once, Never Copy — the registers are read, not duplicated.

• Build on What We Already Have — the registers exist; fix their governance.

• Minimization Is the Privacy Firewall — no shadow hoards to leak.

The architecture state change:

• Today: authoritative registers but no unified model, no stewards, weak quality, some non-compliant data. → Future: one governed data model, accountable stewards, measured quality, a single source per fact, query-not-copy enforced.

The advantages of getting there:

1. A trustworthy substrate—every higher division stands on reliable facts.

2. One consistent truth across the whole state, ending contradictory records.

3. Minimization by construction—no duplicate stores to breach.

4. The precondition that makes cross-agenda composition safe.

5. Legal compliance for data the law already governs.

03 :: The Data Fund

The Layer

The Data Fund division is the live exchange fabric that lets any agency—and any agent—query an authoritative fact from another agency at the moment of decision, so the citizen is asked for nothing the state already knows.

It functions as the circulatory system of the architecture: the channel through which truth moves between registers and agents without ever being copied.

Current state :: what exists now in the Czech Republic

1. The propojený datový fond (PPDF)—the connected data fund, legally enabled as the mechanism for once-only data sharing.

2. ISZR—the information system of the base registers, the reference interface to ROB, ROS, RÚIAN, RPP.

3. eGSB/ISSS—the shared services bus for agency-to-agency data exchange beyond the base registers.

4. The legal basis exists in the register and right-to-digital-services laws; sharing is mandated, not optional.

5. The DIA operates the reference interface as part of its base-register responsibility.

The gap :: where we are slow or fragmented

• Implementation is partial and uneven: many agendas still do not publish or consume data through the fund.

• Real-time, decision-time query is not yet the default; batch and manual exchange persist.

• Without the unified data model and stewardship of Division 02, the fund moves data of uncertain quality.

The agentic target :: 7 properties of the layer done right

1. Real-time — facts are queried at the moment of decision, not synced in batches.

2. Universal — every agenda publishes and consumes through the fund.

3. Once-only by default — the citizen is never asked for a held fact.

4. Purpose-bound — each query is tied to a stated, lawful purpose.

5. Logged — every exchange is recorded and citizen-visible.

6. Minimal — only the needed attribute crosses the fabric.

7. Resilient — the fund is treated as critical infrastructure.

Three patterns of how it works

1. Need → query → deliver

   • An agent needs a fact

   • It queries the fund

   • The authoritative source delivers it

2. Publish → discover → consume

   • Agencies publish authoritative data

   • Consumers discover the source

   • They consume it on demand

3. Bind → log → minimize

   • Each query is purpose-bound

   • Logged for audit

   • Minimized to the needed attribute

4. (patterns held to three)

Ten components :: the building blocks

1. The propojený datový fond (the fabric)

2. ISZR (base-register reference interface)

3. eGSB/ISSS (shared services bus)

4. Real-time query APIs (decision-time access)

5. A service/data catalog (what is available)

6. Purpose-binding (lawful query)

7. Access logging (citizen-visible)

8. Selective disclosure (minimal attributes)

9. Quality gating (no bad data moves)

10. Resilience and availability (critical infrastructure)

The shift :: four “from → to” moves

1. From batch sync to real-time query

   • Facts move at decision time, not on a schedule.

2. From partial to universal participation

   • Every agenda joins the fund.

3. From asking the citizen to asking the source

   • Once-only becomes the default behavior.

4. From opaque to logged, purpose-bound exchange

   • Every movement of data is recorded and constrained.

How to build it

A. Make real-time query the default

• Move agendas from batch and manual exchange to decision-time query through the fund.

• Example: an agent resolving a benefit queries income and address live through ISZR/eGSB rather than requesting documents.

B. Onboard every agenda

• Drive universal publication and consumption so no fact stays trapped in a silo.

• Example: complete the PPDF connection for the agendas behind the first life-event service before launching it.

C. Gate on quality and log on access

• Let no low-quality data move, and make every exchange citizen-visible.

• Example: a citizen dashboard showing which agency queried which fact, for what purpose.

Advantages and risks

Advantages

1. The architectural basis for “ask once, never copy.”

2. Real-time consistency across the whole state.

3. Minimization and auditability by construction.

4. A reusable fabric every life-event service stands on.

Risks

1. A central exchange fabric is a high-value attack target.

2. Real-time availability becomes mission-critical.

3. It moves only data as good as the registers beneath it.

4. Universal onboarding is slow and politically contested.

What needs to be done :: implementation backlog

The work to implement (current → future):

1. Make real-time query the default through ISZR and eGSB/ISSS; retire batch and manual exchange for onboarded agendas.

2. Onboard every agenda to publish and consume through the propojený datový fond.

3. Build a service/data catalog of everything queryable, with its authoritative source.

4. Implement purpose-binding and citizen-visible access logs on every exchange.

5. Add selective disclosure (minimal attribute) at the fabric level.

6. Add quality gating so low-quality data cannot move.

7. Harden the fund as critical infrastructure—availability, redundancy, security.

8. Ship a citizen data-access dashboard showing who queried what, and why.

Principles to apply:

• Ask Once, Never Copy — the fund is the channel that makes once-only real.

• Minimization Is the Privacy Firewall — purpose-bound, minimal, logged.

• The Register Is the Single Source of Truth — the fund moves authoritative facts only.

• The Manual Fallback Never Dies — the fund is treated as critical, with resilience.

The architecture state change:

• Today: the PPDF is legally enabled but partially implemented; batch and manual exchange persist. → Future: universal real-time query, once-only by default, every exchange logged, minimal, and quality-gated, hardened as critical infrastructure.

The advantages of getting there:

1. “Ask once, never copy” becomes architecture, not aspiration.

2. Real-time consistency across every agency.

3. Auditability and privacy built into every data movement.

4. A reusable fabric every life-event service stands on.

5. The citizen sees and controls how their data is used.

04 :: The Agenda Model

The Layer

The Agenda Model division is the machine-readable formalization of what the Czech state may and must do—every agenda, right, obligation, and service—so that an agent can reason over the law itself rather than over a developer’s re-interpretation of it.

It functions as the rulebook of the architecture: the formal model of state authority the agent consults to know what is permitted, required, and owed.

Current state :: what exists now in the Czech Republic

1. RPP (Registr práv a povinností) already encodes the agendas of the state, their legal basis, and the data they may process.

2. RPP defines who may do what to whose data under which agenda—a genuine formal model of authority.

3. The catalog of services under Act 12/2020 enumerates the digital services the state owes citizens.

4. RPP is maintained as a base register under the DIA.

5. It is, today, primarily an access-control and registration instrument, not a reasoning substrate.

The gap :: where we are slow or fragmented

• RPP is underused: it governs access but is not exposed as a model an agent can reason over.

• The catalog of services has lagged its statutory deadlines under Act 12/2020; many services are not yet delivered digitally end-to-end.

• The link between the formal agenda model and executable service logic is weak or absent.

The agentic target :: 7 properties of the layer done right

1. Reasoned-over — the agent consults RPP to know what it may and must do.

2. Complete — every agenda and service is modeled, not just registered.

3. Executable-linked — the model connects to the logic that delivers the service.

4. Authoritative — the model is the single source of “what the state owes.”

5. Versioned — changes in law update the model traceably.

6. Bounded — the agent cannot act outside the modeled agenda.

7. Auditable — every action maps to a modeled right or obligation.

Three patterns of how it works

1. Situation → agenda → permitted action

   • A citizen’s situation is identified

   • The relevant agenda is found in RPP

   • The permitted and required actions follow

2. Model → bound → execute

   • The model defines the bounds

   • The agent acts within them

   • Execution is constrained by the model

3. Law changes → model updates → behavior updates

   • The law changes

   • The model is updated

   • Agent behavior follows automatically

4. (patterns held to three)

Ten components :: the building blocks

1. RPP (the register of rights and obligations)

2. The agenda model (formal authority)

3. The service catalog (what is owed, under Act 12/2020)

4. Legal-basis links (each action to its statute)

5. Executable service logic (the delivery layer)

6. Versioning (law-change traceability)

7. Bounding rules (the agent’s permitted envelope)

8. Obligation triggers (what the state must do, when)

9. Mapping to registers (which data each agenda may use)

10. Audit mapping (action → modeled right)

The shift :: four “from → to” moves

1. From access control to reasoning substrate

   • RPP becomes a model the agent reasons over.

2. From registered to executable

   • The agenda model links to the logic that delivers.

3. From developer interpretation to formal law

   • The agent consults the model, not a coder’s paraphrase.

4. From lagging catalog to delivered services

   • The Act 12/2020 catalog is realized end-to-end.

How to build it

A. Expose RPP as a reasoning model

• Turn the access-control register into a model the agent can query to know its permitted envelope.

• Example: before acting, an agent checks RPP for the agenda, its legal basis, and the data it may use.

B. Close the Act 12/2020 catalog

• Deliver the digital services the law already mandates, end-to-end, starting with the first life event.

• Example: the birth-of-a-child bundle realizes the catalog entries it touches, paying down the statutory backlog.

C. Link the model to executable logic

• Connect each modeled obligation to the service that fulfills it.

• Example: an “obligation to offer” in the model triggers the proactive offer in Division 05.

Advantages and risks

Advantages

1. The agent reasons over the law, not a paraphrase of it.

2. Actions are bounded and auditable against modeled rights.

3. Law changes propagate to behavior traceably.

4. The Act 12/2020 catalog finally gets delivered.

Risks

1. Formalizing the full agenda model is a large undertaking.

2. A wrong model causes wrong action at scale.

3. Law is ambiguous; not everything formalizes cleanly.

4. Over-formalization can ossify discretion the law intends to preserve.

What needs to be done :: implementation backlog

The work to implement (current → future):

1. Expose RPP as a queryable reasoning model—permitted and required actions per agenda, not just access control.

2. Link each modeled agenda to its legal basis and to executable service logic.

3. Implement bounding rules so an agent can act only within its modeled agenda.

4. Add versioning so a change in law updates the model traceably.

5. Encode obligation triggers—what the state must do, and when—as the basis for proactive offers.

6. Map each agenda to the register data it may use.

7. Deliver the Act 12/2020 service catalog end-to-end, starting with the first life event.

8. Build audit mapping—every agent action maps to a modeled right or obligation.

Principles to apply:

• Within Today’s Law First — the model encodes existing law the agent obeys.

• The Official Holds the Pen — the model bounds what the agent may even prepare.

• Predict, Then Offer — obligation triggers are what make proactivity lawful.

• Contestability with a Named Defendant — every action traces to a modeled rule.

The architecture state change:

• Today: RPP is an access-control register, underused; the Act 12/2020 catalog lags its deadlines. → Future: RPP is a reasoning substrate, executable-linked and versioned, and the service catalog is delivered.

The advantages of getting there:

1. The agent reasons over the law itself, not a developer’s paraphrase.

2. Actions are bounded and auditable against modeled rights.

3. Changes in law propagate to behavior traceably.

4. The Act 12/2020 backlog is finally paid down, service by service.

5. Proactive offers rest on a lawful obligation, not a guess.

05 :: The Channels

The Layer

The Channels division is the citizen’s front door to the state—collapsing the many portals, mailboxes, and counters into one continuous conversation that follows the citizen across voice, text, app, and counter.

It functions as the interface of the architecture: the surface where the agentic state meets the human, on the human’s terms.

Current state :: what exists now in the Czech Republic

1. Datové schránky (ISDS)—data mailboxes, now active for businesses and, since 2023, far more individuals.

2. Czech POINT—the network of assisted, in-person counters at post offices and municipalities.

3. Portál občana and Portál veřejné správy—the citizen and public-administration web portals.

4. eDoklady (2024)—the mobile app for identity and documents, growing quickly.

5. Each channel is a separate door the citizen must find, log into, and navigate.

The gap :: where we are slow or fragmented

• The channels are disconnected: the citizen must know which door to use for which need.

• There is no conversational, natural-language entry to the state; everything is forms and portals.

• Uptake lags the infrastructure—DESI 2024 shows strong skills but below-average use of digital public services relative to potential.

The agentic target :: 7 properties of the layer done right

1. Conversational — natural language, not form codes, is the interface.

2. Any-surface — voice, text, app, and counter, interchangeably.

3. Continuous — one conversation that resumes across device and time.

4. Channel-adaptive — the state fits the citizen’s surface, not the reverse.

5. Accessible — voice and plain language include those portals exclude.

6. Assisted-and-self-service — Czech POINT remains a full human path.

7. Identity-bound — every channel ties to the citizen’s authenticated identity.

Three patterns of how it works

1. Speak → understand → act

   • The citizen states a need in natural language

   • The agent understands intent

   • It acts across the state

2. Begin → persist → resume

   • A conversation begins on one surface

   • Context persists

   • It resumes on another

3. Self-service → assisted → human

   • The citizen self-serves where able

   • Assisted where needed

   • A human path always remains

4. (patterns held to three)

Ten components :: the building blocks

1. A conversational agent front end (the new door)

2. Datové schránky (official communication)

3. Czech POINT (assisted human channel)

4. Portál občana (the existing web door)

5. eDoklady (mobile identity and documents)

6. Voice and natural-language understanding

7. Persistent conversation state (continuity)

8. Channel-adaptive rendering (fit the surface)

9. Accessibility features (inclusion)

10. Human handoff (escalation to an official)

The shift :: four “from → to” moves

1. From many doors to one conversation

   • The citizen states a need, not a destination.

2. From forms to natural language

   • The interface is speech and text.

3. From the state’s channel to the citizen’s

   • Access happens on the citizen’s surface and time.

4. From sessions to continuity

   • Context follows the citizen across devices.

How to build it

A. Put a conversational agent in front of the existing portals

• Add a natural-language front door that reaches the existing channels, rather than a sixth portal.

• Example: a citizen says “I’m starting a business” and the agent drives the živnost, tax, and insurance registrations behind the existing systems.

B. Keep Czech POINT as a guaranteed human path

• Preserve the assisted counter as a full route to every agentic service.

• Example: the same birth-of-a-child bundle is completable at a Czech POINT counter, not only in an app.

C. Bind every channel to one identity and one conversation

• Make the conversation continuous across datové schránky, web, app, and counter.

• Example: begin a request in eDoklady and complete it via datová schránka without restarting.

Advantages and risks

Advantages

1. One front door instead of many the citizen must sequence.

2. Natural language includes those portals exclude.

3. Continuity across channels and time.

4. Higher uptake of services the state already offers.

Risks

1. A conversational layer can obscure what the state is doing.

2. Voice and natural language introduce recognition errors.

3. Cross-channel continuity widens the security surface.

4. A new front end must not become yet another disconnected door.

What needs to be done :: implementation backlog

The work to implement (current → future):

1. Build a conversational (voice + text) agent front door over the existing portals—one entry, not a sixth silo.

2. Integrate datové schránky, Portál občana, eDoklady, and Czech POINT into one continuous conversation.

3. Persist conversation and identity context across devices and channels.

4. Add natural-language understanding and voice, with channel-adaptive rendering.

5. Guarantee Czech POINT as a full human path to every agentic service.

6. Build accessibility—plain language and assistive features—as first-class.

7. Implement human handoff to a named official with the full context attached.

8. Bind every channel to the citizen’s authenticated identity.

Principles to apply:

• Any Surface, One Continuous Conversation — the citizen’s channel, not the state’s.

• The Manual Fallback Never Dies — Czech POINT remains a complete route.

• The Citizen Is No Longer the Integrator — one stated need, not many doors.

• Inclusive Legibility — voice and plain language include those portals exclude.

The architecture state change:

• Today: disconnected portals and mailboxes, form-driven, the citizen must find the right door; uptake lags the infrastructure (DESI 2024). → Future: one conversational front door over the existing systems, any surface, continuous, with a guaranteed human path—and higher uptake.

The advantages of getting there:

1. One front door instead of many the citizen must sequence.

2. Natural language includes the digitally excluded.

3. Continuity—context follows the citizen across devices and time.

4. Higher take-up of services the state already offers.

5. The dignity of a state that comes to the citizen.

06 :: The Orchestration Layer

The Layer

The Orchestration Layer is the piece the Czech state does not yet have: the layer that composes agents dynamically across the dozens of agendas to serve a single life event—and the layer whose absence is the reason the foundations underperform.

It functions as the engine of the architecture: the difference between a digital state that holds the pieces and an agentic state that assembles them.

Current state :: what exists now in the Czech Republic

1. Nothing composes across agendas today. Each agenda runs its own system; the citizen is the integrator.

2. The DIA touches 75 agendas and operates 40-plus information systems—but as a portfolio, not an orchestra.

3. The data fund can move facts, but no layer decides which agents to assemble for a given need.

4. Services are per-agenda, not per-life-event.

5. This is the single largest gap between the Czech digital state and an agentic one.

The gap :: where we are slow or fragmented

• There is no orchestration layer at all—it must be built, not improved.

• Cross-agency coordination today is manual, sequential, and citizen-driven.

• Without orchestration, every other division remains a disconnected capability.

The agentic target :: 7 properties of the layer done right

1. Composing — assembles agents across agendas for one request.

2. Dynamic — composition is formed per request, not hard-wired.

3. Life-event-shaped — organized around human moments, not agendas.

4. Discovery-driven — finds the right agents from a capability registry.

5. Verifying — checks every inter-agent handoff (the failure point).

6. Observable — composed services are monitored end-to-end.

7. The moat — the topology of composition is the state’s hardest-won asset.

Three patterns of how it works

1. Life event → decompose → compose

   • A life event is declared

   • It is decomposed into agenda tasks

   • The relevant agents are composed

2. Registry → discover → assemble

   • Agents register capabilities

   • The orchestrator discovers them

   • It assembles them dynamically

3. Handoff → verify → continue

   • One agent hands off to another

   • The handoff is verified

   • The composition continues safely

4. (patterns held to three)

Ten components :: the building blocks

1. The orchestrator (the composer)

2. An agent capability registry (what each agent does)

3. Life-event-to-agenda mappings (the bundles)

4. Discovery and routing (find the right agents)

5. Inter-agent protocols (how agents hand off)

6. Handoff verification (the critical check)

7. The data fund (the shared substrate)

8. Composition policies (which agents may compose)

9. End-to-end monitoring (observe the whole)

10. Versioning (replace agents without breaking the whole)

The shift :: four “from → to” moves

1. From portfolio to orchestra

   • The state’s systems are composed, not merely owned.

2. From per-agenda to per-life-event

   • Service is organized around human moments.

3. From citizen-driven to state-driven coordination

   • The orchestrator integrates, not the citizen.

4. From static integration to dynamic composition

   • Compositions form per request.

How to build it

A. Build the orchestrator as the program’s core

• Treat the orchestration layer, not another portal, as the central deliverable.

• Example: the DIA builds one orchestrator that composes its 75 agendas, rather than 75 disconnected apps.

B. Start with one life-event composition

• Prove the layer on a single bundle before generalizing.

• Example: the birth-of-a-child service is the first composition—registry of birth triggers health, social, and benefit agents together.

C. Verify every handoff

• Make inter-agent handoff verification a first-class requirement, since it is where agentic systems fail.

• Example: the flood-response composition verifies that the insurance, housing, and permit steps each completed before reporting success.

Advantages and risks

Advantages

1. The missing layer that makes the state genuinely agentic.

2. Cross-agency services that match real life.

3. A compounding orchestration moat unique to the state that builds it.

4. Reuse—new life events recombine existing agents.

Risks

1. It concentrates enormous power and failure in one layer.

2. Inter-agent handoffs are the most common agentic failure mode.

3. It requires cross-agenda cooperation, which is politically hard.

4. A composed service is only as reliable as its weakest agent.

What needs to be done :: implementation backlog

The work to implement (current → future):

1. Build the orchestrator as the program’s central deliverable—not another portal.

2. Build an agent capability registry—what each of the agencies’ agents can do.

3. Define life-event → agenda bundles (birth, job loss, business, flood).

4. Implement discovery, routing, and inter-agent protocols for dynamic composition.

5. Implement handoff verification on every boundary—the agentic failure point.

6. Add end-to-end monitoring of each composed service.

7. Add composition policies governing which agents may compose.

8. Add versioning so agents are replaced without breaking compositions.

9. Ship the first composition (birth-of-a-child) and generalize from it.

Principles to apply:

• Agents Compose Across Ministries — composition is the product and the moat.

• The Life Event Is the Unit of Service — bundles, not agendas.

• The Citizen Is No Longer the Integrator — the orchestrator integrates, not the human.

• Resilient Pluralism — verified handoffs and modular, replaceable agents.

The architecture state change:

• Today: a portfolio of 75 agendas and 40-plus information systems with no layer composing them; the citizen is the integrator. → Future: an orchestra—dynamic, per-life-event composition across agendas, with verified handoffs and the state doing the integration.

The advantages of getting there:

1. The single missing layer that makes the state genuinely agentic.

2. Cross-agency services that match how people actually live.

3. A compounding orchestration moat unique to the state that builds it.

4. Reuse—new life events recombine existing agents.

5. The end of the citizen as the state’s unpaid clerk.

07 :: The Runtime

The Layer

The Runtime division is the model cognition the state runs on—the inspectable, EU-hostable, AI-Act-conformant models that power every agent—so that what the Czech state’s institutions may conclude and say is owned and auditable, not rented opaquely from a foreign power.

It functions as the mind of the architecture: the layer that reasons, drafts, and decides-in-preparation, beneath the orchestration that directs it.

Current state :: what exists now in the Czech Republic

1. The Czech state runs no production AI agents at scale; this layer is largely greenfield.

2. It has a National AI Strategy 2030 (approved 2024) and a 2026 action component within Digitální Česko.

3. A draft AI implementation law (2025) and the EU AI Act (2024/1689) define the coming legal frame.

4. There is no sovereign model-hosting capability dedicated to public administration.

5. Public bodies experiment with foreign commercial models, mostly outside any sovereign, inspectable runtime.

The gap :: where we are slow or fragmented

• There is no sovereign, inspectable runtime the state owns and can audit.

• AI use is ad hoc and ungoverned, risking exactly the vendor-dependence the agentic state must avoid.

• The legal frame (AI Act transposition, draft law) is arriving, not yet in force.

The agentic target :: 7 properties of the layer done right

1. Sovereign — the state owns and can replace its models.

2. Inspectable — open-weight or auditable, not an opaque API for core decisions.

3. EU-hostable — run on infrastructure the state controls.

4. AI-Act-conformant — the high-risk regime met by construction.

5. Bounded — the runtime acts only within the agenda model and approval layer.

6. Observable — every inference is logged for audit.

7. Revocable — no punitive lock-in to any single provider.

Three patterns of how it works

1. Host → audit → govern

   • The state hosts inspectable models

   • It audits their behavior

   • It governs what they may do

2. Conform → certify → deploy

   • Systems are built AI-Act-conformant

   • Certified

   • Deployed into agents

3. Reason → prepare → defer

   • The runtime reasons over a case

   • Prepares a determination

   • Defers the rights decision to the official

4. (patterns held to three)

Ten components :: the building blocks

1. Inspectable, EU-hostable models (the cognition)

2. Sovereign inference infrastructure (where they run)

3. AI-Act conformance (the high-risk regime)

4. Model audit and red-teaming (inspection)

5. The agenda model (the bounds on reasoning)

6. Inference logging (audit)

7. Evaluation and quality gates (is it good enough)

8. Provenance and bills-of-materials (supply-chain integrity)

9. A national capability (skills to run and modify)

10. A revocability plan (no lock-in)

The shift :: four “from → to” moves

1. From rented to sovereign cognition

   • The state controls the model that reasons for it.

2. From opaque to inspectable

   • Behavior is auditable and modifiable.

3. From ad hoc to governed AI

   • Use is bounded, conformant, and logged.

4. From lock-in to revocability

   • Dependence is always reversible.

How to build it

A. Stand up a sovereign, inspectable runtime

• Host EU-hostable models the state can audit, for any rights-relevant reasoning.

• Example: run the official-facing agent of Division 09 on an inspectable model under the EU AI Act, not an opaque foreign API.

B. Make AI-Act conformance the build standard

• Treat the high-risk regime as the design baseline, turning compliance into trust.

• Example: the draft Czech implementing law and the AI Act define the conformance the runtime meets from day one.

C. Bound the runtime by the agenda model

• Let the runtime reason only within what RPP permits, and defer rights decisions upward.

• Example: the model prepares a benefit determination but cannot issue it—the official does.

Advantages and risks

Advantages

1. The state’s institutions decide what they may conclude and say.

2. Auditable, governable cognition.

3. Conformance turned into trust.

4. A sovereign capability exportable to other EU states.

Risks

1. Sovereign hosting is costlier and slower than renting frontier APIs.

2. EU-hostable models may trail the global frontier in capability.

3. Compute and chip access remain partly externally constrained.

4. Building the capability requires scarce skills and sustained funding.

What needs to be done :: implementation backlog

The work to implement (current → future):

1. Stand up a sovereign, EU-hostable, inspectable model runtime for public administration.

2. Build AI-Act conformance into the runtime (the high-risk regime) by construction.

3. Bound the runtime by the agenda model and the approval layer—it reasons, it does not decide rights.

4. Implement inference logging and evaluation/quality gates.

5. Add provenance and AI bills-of-materials, with continuous red-teaming for poisoning.

6. Build national capability—the skills to run, evaluate, and modify models.

7. Define a revocability and exit plan—no punitive lock-in to any provider.

8. Transpose the EU AI Act and pass the implementing law as the legal base.

Principles to apply:

• Sovereign-European Runtime by Construction — owned, inspectable cognition.

• Within Today’s Law First — start on internal, official-facing use.

• The Official Holds the Pen — the runtime defers every rights decision.

• Resilient Pluralism — diversity, provenance, fail-soft from day one.

The architecture state change:

• Today: no production agents at scale, ad hoc use of foreign commercial models, ungoverned. → Future: a sovereign, inspectable, AI-Act-conformant runtime the state owns, governs, logs, and can revoke.

The advantages of getting there:

1. The state’s institutions decide what they may conclude and say.

2. Auditable, governable cognition rather than a rented black box.

3. AI-Act conformance turned into a trust asset.

4. A sovereign capability exportable to other EU states.

5. No vendor lock-in—models are replaceable.

08 :: The Approval Layer

The Layer

The Approval Layer is the human decision on a citizen’s rights: the agent prepares the case, a named official approves it, and the decision-maker of record remains the official—exactly as the správní řád requires today—so the whole architecture runs inside existing law.

It functions as the keystone of the architecture: the layer that makes everything beneath it lawful and accountable without changing a statute.

Current state :: what exists now in the Czech Republic

1. The správní řád (Administrative Procedure Code) requires that a human official decide matters affecting rights.

2. Officials today decide manually, assembling cases by hand across systems.

3. There is no agent-to-official preparation workflow; the agent does not yet exist to prepare.

4. The legal principle—a human decides—is exactly what lets the agentic state start without new law.

5. Accountability for a decision already attaches to a named official, a property to preserve.

The gap :: where we are slow or fragmented

• Officials spend effort on case assembly that an agent could do.

• There is no tooling for an official to review and approve an agent-prepared case.

• Without designed approval, automation risks degrading into rubber-stamping.

The agentic target :: 7 properties of the layer done right

1. Preparation, not decision — the agent readies the case; the official decides.

2. Within existing law — preserving the official as decider needs no statute change.

3. Empowered — the official has reasons, the power to amend, and the time to use them.

4. Named — accountability attaches to a specific human.

5. Bounded — routine actions complete autonomously; rights decisions route to a human.

6. Measured — override rates prove the approval is real, not perfunctory.

7. Auditable — preparation and approval are distinct, logged events.

Three patterns of how it works

1. Prepare → present → approve

   • The agent prepares the case

   • Presents it to the official

   • The official approves, amends, or rejects

2. Routine → autonomous; rights → human

   • Routine actions complete autonomously

   • Rights decisions route to a human

   • The boundary governs which path applies

3. Draft → review → own

   • The agent drafts

   • The official substantively reviews

   • The official owns the decision

4. (patterns held to three)

Ten components :: the building blocks

1. A preparation engine (agent casework)

2. An approval interface (where the official decides)

3. The rights/routine boundary (what needs a human)

4. Reason traces (so review is meaningful)

5. Amendment capacity (the official can change the draft)

6. Named accountability (the human owner)

7. Override metrics (proof the official decides)

8. Audit logs (preparation and approval as distinct events)

9. Escalation paths (complex cases upward)

10. Official training (supervisors of agent casework)

The shift :: four “from → to” moves

1. From manual assembly to agent preparation

   • The agent readies the case; the official decides.

2. From new law to existing law

   • Preserving the official as decider keeps the správní řád intact.

3. From rubber-stamp to empowered approval

   • The official is informed and able to change the outcome.

4. From diffuse to named accountability

   • Each decision has a human owner of record.

How to build it

A. Draw the rights/routine boundary explicitly

• Define which actions an agent may complete and which it may only prepare.

• Example: an agent may file a notification autonomously but only prepares a benefit determination for the official.

B. Build the approval interface

• Give officials the reasons, the power to amend, and the time to decide—then measure overrides.

• Example: the Czech concept’s own principle—”AI připravuje podklady, úředník schvaluje, proto není potřeba měnit správní řád”—made into a working tool.

C. Keep preparation and approval as separate logged acts

• Record what the agent prepared and what the official decided, distinctly.

• Example: an audit trail showing the draft and the human decision as two events.

Advantages and risks

Advantages

1. Due process preserved in substance.

2. Deployable within existing law—no statute change.

3. A named, accountable human for every rights decision.

4. Officials freed from case assembly for judgment.

Risks

1. Automation bias can hollow approval into rubber-stamping.

2. Case volume can pressure officials toward perfunctory review.

3. The rights/routine boundary will be contested at the edges.

4. Without real override capacity, approval becomes theater.

What needs to be done :: implementation backlog

The work to implement (current → future):

1. Define the rights/routine boundary explicitly—what an agent may complete versus only prepare.

2. Build the agent preparation engine—automated casework assembly across systems.

3. Build the official approval interface—reasons, the power to amend, and the time to use them.

4. Implement override metrics and audit, recording preparation and approval as distinct events.

5. Preserve named accountability for every rights decision.

6. Build escalation paths for complex cases.

7. Train officials as supervisors of agent casework, not form-fillers.

8. Stay within the správní řád—no statutory change required for phase one.

Principles to apply:

• The Official Holds the Pen — the keystone that keeps the system lawful.

• Within Today’s Law First — preservation of the official as decider needs no new law.

• Augmentation Over Automation — an empowered approver, not a rubber stamp.

• Contestability with a Named Defendant — accountability has a human address.

The architecture state change:

• Today: officials assemble cases by hand; no agent-to-official workflow; accountability already attaches to the official. → Future: the agent prepares, the official approves with real power to amend, overrides are measured, and the whole flow runs inside existing law.

The advantages of getting there:

1. Due process preserved in substance, not just form.

2. Deployable now—no statute must move first.

3. Officials freed from case assembly for genuine judgment.

4. A named, accountable human for every rights decision.

5. The crumple zone refused by design.

09 :: Decision and Audit

The Layer

The Decision and Audit division ensures that every action in the agentic state produces a reason and a record, and that every citizen can contest it—holding to the European principle that a computation which determines an outcome is itself the regulated decision.

It functions as the conscience of the architecture: the layer that makes power answerable and harm reversible.

Current state :: what exists now in the Czech Republic

1. Administrative decisions today produce a file and a written justification under the správní řád.

2. Appeal rights exist through established administrative and judicial review.

3. There are no machine reason traces for automated steps, because there are no agents yet.

4. Logging exists for systems, but not as citizen-facing, decision-level reasons.

5. The legal frame for automated decisions is arriving via the AI Act and the draft implementing law.

The gap :: where we are slow or fragmented

• There is no reason-trace or contestability layer designed for agentic decisions.

• Existing appeal is slow and document-heavy, ill-suited to high-volume automated steps.

• Without this layer, automated preparation risks harm without a clear, fast remedy.

The agentic target :: 7 properties of the layer done right

1. Reasoned — every decision carries an inspectable reason.

2. Recorded — decisions are logged and reproducible.

3. Contestable — an affordable appeal reaches an accountable human.

4. SCHUFA-aligned — the determining computation is treated as the decision.

5. Assisted — the citizen gets help to understand and challenge.

6. Time-bound — remedies arrive on a guaranteed timeline.

7. Transparent — citizens can see what was decided and why.

Three patterns of how it works

1. Decide → explain → contest

   • A decision is made

   • Its reasons are produced

   • The citizen can contest it

2. Appeal → human review → remedy

   • An appeal is filed affordably

   • An accountable human reviews

   • A remedy issues where warranted

3. Determine → regulate → assign

   • The determining computation is identified

   • Regulated as the decision

   • A named human is assigned responsibility

4. (patterns held to three)

Ten components :: the building blocks

1. Reason traces (for every decision)

2. Reproducible decision logs

3. Affordable appeal channels

4. An accountable human reviewer

5. The SCHUFA principle (determining computation = decision)

6. Pre-assigned liability (a named defendant)

7. Public-option assistance (to contest)

8. Explainability standards

9. Independent administrative and judicial review

10. Time-bound remedy guarantees

The shift :: four “from → to” moves

1. From file to reason trace

   • Every decision carries an inspectable reason.

2. From slow appeal to time-bound remedy

   • Recourse is fast, affordable, and effective.

3. From rubber-stamp to regulated computation

   • The determining computation is the decision.

4. From opaque to transparent

   • Citizens see what was decided and why.

How to build it

A. Generate a reason for every decision

• Make an inspectable reason trace a hard requirement of every agent action.

• Example: a benefit determination arrives with the facts queried, the rule applied, and the official who approved it.

B. Build a fast, assisted appeal

• Add an affordable, assisted appeal designed for high-volume automated steps.

• Example: a public-option agent that explains a decision and prepares the appeal, so contestability is not a privilege.

C. Adopt the SCHUFA principle in practice

• Treat any computation that effectively determines an outcome as the regulated decision.

• Example: the CJEU SCHUFA ruling (C-634/21) as the governing precedent, reaching past the stamp to the model.

Advantages and risks

Advantages

1. Power made answerable and harm reversible.

2. A named defendant and a fast remedy for every decision.

3. Trust earned through transparency and contestability.

4. Governable, auditable decision-making.

Risks

1. Reason traces can be gamed or made uninformative.

2. Appeal volume can overwhelm capacity without careful design.

3. Explainability of complex models is technically hard.

4. Assistance to contest requires sustained funding to stay real.

What needs to be done :: implementation backlog

The work to implement (current → future):

1. Mandate reason traces for every agent decision and preparation.

2. Build reproducible decision logs, reconstructable for review.

3. Build a fast, affordable, assisted appeal designed for high-volume automated steps.

4. Implement public-option assistance—an agent that explains a decision and prepares the appeal.

5. Adopt the SCHUFA principle in practice—the determining computation is the regulated decision.

6. Pre-assign liability—a named defendant—before any deployment.

7. Set explainability standards and time-bound remedy guarantees.

8. Provide citizen-visible transparency of what was decided and why.

Principles to apply:

• Contestability with a Named Defendant — always an answer, always a defendant.

• Augmentation Over Automation — a human reviews the appeal.

• Minimization Is the Privacy Firewall — logs are purpose-bound and minimal.

• Inclusive Legibility — assistance makes contestability available to all.

The architecture state change:

• Today: administrative files and written justifications, slow document-heavy appeal, no machine reasons for automated steps. → Future: a reason trace for every decision, a fast assisted appeal, the SCHUFA principle enforced, a named defendant, and time-bound remedies.

The advantages of getting there:

1. Power made answerable and harm reversible.

2. A named defendant and a fast remedy for every decision.

3. Trust earned through transparency and contestability.

4. Governable, auditable decision-making.

5. Contestability that is not a privilege of the well-resourced.

10 :: Governance and Mandate

The Layer

The Governance and Mandate division is who owns and authorizes the agentic state: the DIA as delivery owner, Act 12/2020 and the Digitální Česko program as the existing frame, and a founding government resolution as the act that turns a fragmented mandate into an engine.

It functions as the will of the architecture: the layer that decides the agentic state shall exist, and holds someone accountable for building it.

Current state :: what exists now in the Czech Republic

1. The DIA (since 2023) is the central authority for the digital agenda, gestor of 21 agendas and active in 54 more.

2. Act 12/2020 Sb. on the right to digital services is the country’s “digital constitution,” mandating a service catalog.

3. The Digitální Česko program and its 2026 implementation plan set direction and budget.

4. The National AI Strategy 2030 and the draft AI law frame the AI dimension.

5. Yet delivery is fragmented, and Act 12/2020’s catalog has lagged its statutory deadlines.

The gap :: where we are slow or fragmented

• Ownership exists but delivery is fragmented across bodies and budgets.

• The legal mandate (Act 12/2020) outran execution; deadlines slipped.

• There is no single founding act that names the agentic-state goal and an accountable owner.

The agentic target :: 7 properties of the layer done right

1. Mandated — a government resolution defines the goal and assigns responsibility.

2. Owned — the DIA is the named delivery owner.

3. Capable — an expert AI center gives the mandate hands.

4. Within law — phase one runs under existing legislation.

5. Accountable — public milestones make progress visible.

6. Resourced — a budget line backs the mandate.

7. Adjustable — the mandate updates as evidence arrives.

Three patterns of how it works

1. Resolution → owner → mobilization

   • A resolution issues

   • It assigns the owner

   • The administration mobilizes

2. Mandate → milestones → accountability

   • The goal is mandated

   • Milestones are set

   • Progress is held accountable

3. Pilot → evidence → targeted law

   • Pilots run within existing law

   • They produce evidence

   • Targeted legislation follows

4. (patterns held to three)

Ten components :: the building blocks

1. A founding government resolution (the mandate)

2. The DIA (delivery owner)

3. An expert AI center (capability)

4. Act 12/2020 (the right to digital services)

5. Digitální Česko (program and budget)

6. The National AI Strategy 2030 (AI direction)

7. Public milestones (accountability)

8. A budget line (resourcing)

9. The reform backlog (evidence-based law)

10. A revision mechanism (adjust to evidence)

The shift :: four “from → to” moves

1. From fragmented delivery to a named owner

   • The DIA owns the transformation end-to-end.

2. From lagging catalog to delivered services

   • Act 12/2020’s mandate is realized via pilots.

3. From statute-first to resolution-first

   • The founding is an executive act that starts now.

4. From ambition to accountable milestones

   • Progress is public and measurable.

How to build it

A. Found it by resolution and name the DIA

• Issue a government resolution defining the goal and assigning the DIA as owner.

• Example: a usnesení vlády that mandates the agentic-state program and an expert AI center at the DIA.

B. Deliver Act 12/2020 through the program

• Use the agentic program to finally realize the lagging service catalog.

• Example: each life-event bundle pays down the statutory catalog backlog it touches.

C. Legislate from evidence

• Write the targeted implementing law from working pilots, aligned with the AI Act.

• Example: the draft Czech AI law follows the pilots, enabling the next phase rather than blocking the first.

Advantages and risks

Advantages

1. A clear owner and a fast, executive founding.

2. The existing legal frame realized, not duplicated.

3. Accountable, public milestones.

4. Evidence-based, targeted legislation.

Risks

1. A resolution lacks the durability of law and can be reversed.

2. Fragmented budgets can still starve delivery.

3. Mandate without capability is empty—the expert center must be real.

4. Political turnover can break continuity across phases.

What needs to be done :: implementation backlog

The work to implement (current → future):

1. Issue a founding government resolution (usnesení vlády) naming the goal and the DIA as owner.

2. Stand up an expert AI center at the DIA as the delivery and capability home.

3. Set public milestones and a dedicated budget line.

4. Use the program to deliver the lagging Act 12/2020 service catalog.

5. Maintain an evidence-based reform backlog—the few laws the next phase needs.

6. Establish a reporting cadence and a revision mechanism.

7. Align with the National AI Strategy 2030 and Digitální Česko 2026.

8. Secure cross-ministry authority to orchestrate across agendas.

Principles to apply:

• Found It by Resolution, Not Statute — a fast executive mandate, not a new law.

• Officials Before Citizens — the sequencing the mandate enforces.

• Within Today’s Law First — phase one runs under existing legislation.

• Build on What We Already Have — the DIA, Digitální Česko, and Act 12/2020 exist.

The architecture state change:

• Today: ownership exists in the DIA, but delivery is fragmented and the Act 12/2020 catalog has slipped its deadlines. → Future: a single founding resolution, a capable owner (the DIA plus an expert center), public milestones, a delivered catalog, and evidence-based law.

The advantages of getting there:

1. A fast, executive founding that can start in weeks, not years.

2. A clear, accountable owner of the transformation.

3. The existing legal frame realized rather than duplicated.

4. Public, measurable milestones.

5. Targeted, evidence-based legislation instead of speculation.

11 :: Infrastructure

The Layer

The Infrastructure division is the compute and hosting floor beneath the cognition—the eGovernment cloud, the CLOUDIA private cloud, and the data centers on which the sovereign runtime must stand—because a state that cannot host its own cognition does not truly own it.

It functions as the ground of the architecture: the physical and operational floor that makes sovereign cognition possible.

Current state :: what exists now in the Czech Republic

1. The eGovernment cloud (eGC) is the framework for hosting public information systems.

2. CLOUDIA, the DIA’s private cloud, hosts part of the state’s systems.

3. Other systems run in state data centers and, in part, commercial clouds.

4. DESI 2024 notes a planned digital-transformation budget on the order of EUR 1.77 billion (about 0.6% of GDP).

5. There is no infrastructure dedicated to hosting a sovereign model runtime at scale.

The gap :: where we are slow or fragmented

• Hosting is split across eGC, CLOUDIA, state and commercial clouds, without a sovereign runtime floor.

• There is no strategic compute capacity earmarked for public-administration AI.

• Dependence on commercial clouds for AI risks the vendor lock-in the agentic state must avoid.

The agentic target :: 7 properties of the layer done right

1. Sovereign — compute and hosting the state controls.

2. Sufficient — capacity sized to run core governance under load.

3. Inspectable-friendly — able to host open, auditable models.

4. Resilient — redundant, with no single point of failure.

5. Consolidated — a coherent floor, not a scattered patchwork.

6. Efficient — cost-managed against real governance load.

7. Trusted-partner-extensible — able to share sovereign capacity across the EU.

Three patterns of how it works

1. Host → run → scale

   • The floor hosts the runtime

   • Agents run on it

   • Capacity scales with load

2. Consolidate → secure → operate

   • Scattered hosting is consolidated

   • Secured as critical infrastructure

   • Operated reliably

3. Reserve → provision → sustain

   • Strategic compute is reserved

   • Provisioned to agents

   • Sustained under stress

4. (patterns held to three)

Ten components :: the building blocks

1. The eGovernment cloud (eGC)

2. CLOUDIA (the DIA private cloud)

3. State data centers

4. Sovereign inference capacity (the missing piece)

5. Redundancy and resilience (no single point of failure)

6. Security and isolation (critical-infrastructure grade)

7. Capacity planning (sized to load)

8. Cost management (efficiency)

9. Trusted-partner capacity (EU sharing)

10. Energy and continuity (the power floor)

The shift :: four “from → to” moves

1. From scattered hosting to a consolidated floor

   • A coherent sovereign base, not a patchwork.

2. From commercial dependence to sovereign capacity

   • The state can host its own cognition.

3. From general cloud to inference-ready capacity

   • Capacity sized and shaped for agentic load.

4. From single-sourced to resilient and shared

   • Redundant and extensible across trusted partners.

How to build it

A. Consolidate onto a sovereign floor

• Bring AI hosting onto eGC/CLOUDIA-based sovereign capacity rather than scattered commercial clouds.

• Example: host the inspectable runtime of Division 07 on sovereign infrastructure the DIA controls.

B. Reserve strategic compute

• Earmark inference capacity sufficient to run core governance under load.

• Example: a compute reserve sized to the life-event services in the rollout, with headroom.

C. Build resilience and trusted-partner extensibility

• Make the floor redundant and able to share capacity across EU partners.

• Example: a trusted-partner arrangement for surge capacity that keeps cognition sovereign.

Advantages and risks

Advantages

1. Cognition the state genuinely owns and can audit.

2. A consolidated, resilient hosting floor.

3. Capacity sized to real governance load.

4. Extensibility across trusted EU partners.

Risks

1. Sovereign capacity is costlier than commercial cloud.

2. Building inference capacity requires scarce skills and chips.

3. Consolidation is a large migration with its own risk.

4. Underestimating load leaves the floor unable to sustain governance.

What needs to be done :: implementation backlog

The work to implement (current → future):

1. Consolidate AI hosting onto sovereign eGC/CLOUDIA-based capacity, off scattered commercial clouds.

2. Build or reserve strategic inference capacity sized to the life-event services, with headroom.

3. Make the floor inspectable-model-friendly—able to host open, auditable models.

4. Build redundancy and critical-infrastructure-grade security—no single point of failure.

5. Plan capacity against real governance load and manage cost.

6. Establish trusted-partner extensibility for surge capacity that keeps cognition sovereign.

7. Address the power and continuity floor beneath the compute.

8. Migrate scattered hosting onto the consolidated sovereign floor.

Principles to apply:

• Sovereign-European Runtime by Construction — the compute floor makes ownership real.

• Resilient Pluralism — redundancy and no single point of failure.

• Build on What We Already Have — eGC and CLOUDIA are the starting point.

• The Manual Fallback Never Dies — the floor is critical infrastructure.

The architecture state change:

• Today: hosting split across eGC, CLOUDIA, state data centers, and commercial clouds, with no sovereign inference floor. → Future: a consolidated, sovereign, inference-ready, resilient floor with a strategic compute reserve, extensible across trusted EU partners.

The advantages of getting there:

1. Cognition the state genuinely owns and can audit.

2. A consolidated, resilient hosting floor instead of a patchwork.

3. Capacity sized to real governance load.

4. Extensibility across trusted EU partners without losing sovereignty.

5. Lock-in avoided at the infrastructure layer.

12 :: Resilience

The Layer

The Resilience division guarantees that the agentic state, in gaining coherence, does not lose the accidental robustness of its fragmented past—preserving the manual fallback, model diversity, data stewardship, and data quality that keep the whole architecture trustworthy.

It functions as the immune system of the architecture: the layer that keeps a more powerful, more coupled state from becoming a more brittle one.

Current state :: what exists now in the Czech Republic

1. The current state is fragmented—and therefore fails locally, one agenda at a time.

2. Manual procedures still exist everywhere; the state can be, and is, run by hand.

3. But the Informační koncepce admits weak data quality, no unified model, and missing data stewards.

4. There is no agentic monoculture yet—and no guarantee one will not form carelessly.

5. Czech POINT and human offices provide a real, universal non-digital path today.

The gap :: where we are slow or fragmented

• Data quality and stewardship are weak, undermining trust in any automation built on them.

• As the state consolidates onto an orchestration layer and a shared runtime, it risks trading local failure for correlated failure.

• There is no designed manual-fallback guarantee for the agentic flows that do not yet exist.

The agentic target :: 7 properties of the layer done right

1. Manual-fallback-guaranteed — a human path always exists, by design.

2. Diverse — multiple models and vendors, no monoculture.

3. Fail-soft — services degrade gracefully and revert to humans.

4. Stewarded — data has accountable owners (closing Division 02’s gap).

5. Quality-monitored — systematic, continuous data-quality measurement.

6. Provenance-checked — models carry bills-of-materials; poisoning is caught.

7. Inclusive — those who cannot use agents are first-class citizens.

Three patterns of how it works

1. Fail → degrade → revert

   • A component fails

   • Services degrade gracefully

   • Core functions revert to human procedure

2. Diversify → isolate → contain

   • Multiple models and vendors are deployed

   • Critical subsystems isolated

   • Failures contained locally

3. Steward → measure → trust

   • Data is stewarded

   • Quality is measured

   • The architecture earns trust

4. (patterns held to three)

Ten components :: the building blocks

1. A guaranteed non-digital path (Czech POINT and beyond)

2. Preserved manual procedures (run by hand)

3. Model and vendor diversity (no monoculture)

4. Fail-soft architectures (graceful degradation)

5. Data stewards (accountable owners)

6. Data-quality monitoring (systematic)

7. Provenance and bills-of-materials (poisoning defense)

8. Decoupled critical subsystems (no cascade)

9. Institutional memory (the knowledge to run by hand)

10. Independent resilience audits

The shift :: four “from → to” moves

1. From accidental to engineered resilience

   • Robustness is designed, not a byproduct of fragmentation.

2. From weak to stewarded, quality-monitored data

   • The architecture’s trust is built on reliable facts.

3. From monoculture risk to diversity

   • No single failure takes the whole state down.

4. From digital-only to a guaranteed human path

   • No citizen is left without recourse.

How to build it

A. Fix data quality and stewardship first

• Close the Informační koncepce’s own diagnosed gaps before scaling automation on top.

• Example: appoint data stewards and stand up quality monitoring as a precondition of the first life-event service.

B. Guarantee the manual fallback in law and design

• Ensure every agentic service has a full human equivalent, and keep the staff who can run it.

• Example: the birth-of-a-child bundle remains completable at a Czech POINT counter, with trained staff.

C. Refuse the monoculture

• Mandate model and vendor diversity, provenance, and fail-soft design for critical functions.

• Example: when one model is quarantined for suspected poisoning, services revert to a second model or to human procedure without interruption.

Advantages and risks

Advantages

1. Coherence gained without brittleness imported.

2. Trustworthy automation built on stewarded, quality data.

3. No outage, failure, or exclusion leaves a citizen without recourse.

4. Resilience against monoculture and poisoning.

Risks

1. Maintaining manual capacity consumes resources that look idle.

2. Diversity raises integration cost and complexity.

3. Fixing data quality is slow and unglamorous, and may be skipped.

4. Preserved fallbacks can atrophy in practice if not exercised.

What needs to be done :: implementation backlog

The work to implement (current → future):

1. Fix data quality and stewardship first—the precondition for trusting anything built on top.

2. Guarantee a non-digital path in law and design for every agentic service.

3. Keep trained staff and documented manual procedures—the capacity to run the state by hand.

4. Mandate model and vendor diversity—refuse the monoculture for critical functions.

5. Build fail-soft architectures—revert to a second model or to human procedure on failure.

6. Implement provenance and AI bills-of-materials, with continuous poisoning red-teams.

7. Decouple critical subsystems so a failure cannot cascade.

8. Run independent resilience audits and exercise the fallback regularly so it does not atrophy.

Principles to apply:

• The Manual Fallback Never Dies — a human path always exists.

• Resilient Pluralism — diversity, provenance, and fail-soft design.

• The Register Is the Single Source of Truth — stewarded, quality-monitored data.

• Inclusive Legibility — those who cannot use agents are first-class citizens.

The architecture state change:

• Today: accidental resilience from fragmentation, but weak data quality, missing stewards, and no designed fallback for agentic flows. → Future: engineered resilience—stewarded, quality-monitored data, a guaranteed manual fallback, model and vendor diversity, fail-soft design, and provenance.

The advantages of getting there:

1. Coherence gained without importing brittleness.

2. Automation built on stewarded, trustworthy data.

3. No outage, failure, or exclusion leaves a citizen without recourse.

4. Resistance to monoculture failure and model poisoning.

5. The whole architecture earns, and keeps, public trust.

Action plan :: building the Czech Agentic State

The twelve divisions describe an architecture, not a wish. This plan sequences them onto the existing Czech stack and the realistic, within-the-law path—foundations first, then the missing orchestration layer, then life events, then guarantees and law—each step tagged to the divisions it builds. It closes with a named deliverable.

Phase 0 :: Mandate and foundations (Q3 2026)

1. Found the program by government resolution, naming the DIA as owner and standing up an expert AI center (Division 10).

2. Fix data quality and stewardship—appoint data stewards, build the unified data model, start quality monitoring (Divisions 02, 12).

3. Converge identity onto the EU wallet and design agent identity (Division 01).

Phase 1 :: Officials first, on a sovereign runtime (Q3 2026 – 2027)

4. Deploy an internal official-facing agent on one ministry, on an inspectable, AI-Act-conformant runtime (Divisions 07, 09).

5. Build the approval workflow—the agent prepares, the official approves, within the správní řád (Division 08).

6. Expose RPP as a reasoning model so the agent acts only within the modeled agenda (Division 04).

Phase 2 :: The orchestration layer and the first life event (2027 – 2028)

7. Build the orchestration layer—the missing engine—as the program’s core deliverable (Division 06).

8. Ship the birth-of-a-child service end-to-end, composing health, social, and benefit agents (Divisions 05, 06).

9. Make once-only real by querying the propojený datový fond live, never copying (Divisions 03, 02).

10. Add the channels—a conversational front door over the existing portals, with Czech POINT as the guaranteed human path (Divisions 05, 12).

Phase 3 :: Generalize, harden, legislate (2028 →)

11. Extend to job loss, business, and disaster recovery, reusing the orchestration spine (Divisions 06, 04).

12. Consolidate the sovereign compute floor under the runtime (Division 11).

13. Guarantee contestability and the manual fallback, and refuse the monoculture (Divisions 09, 12).

14. Legislate from evidence—the targeted Czech AI implementing law, aligned with the EU AI Act and the identity wallet (Divisions 10, 07).

Deliverable :: The Czech Agentic State Architecture Charter

A single governing artifact that commits the Czech Republic to the twelve-division architecture and specifies, for each division, its current baseline (the real systems—ROB, ROS, RÚIAN, RPP, NIA, BankID, eDoklady, datové schránky, the propojený datový fond, ISZR, eGSB/ISSS, eGC, CLOUDIA, the DIA), its diagnosed gap (fragmented identity, missing data model and stewards, partial data-fund implementation, an absent orchestration layer, no sovereign runtime, a lagging Act 12/2020 catalog), and its agentic target with the owner, milestones, and the within-the-law path to reach it.

The Czech Republic does not need to invent the agentic state. It needs to build one missing layer—orchestration—on foundations it already has, fix the data beneath them, keep a human holding the pen, and start within its own law. Done in that order, Czechia becomes not the country that bought the most government AI, but the first agentic state in Europe that is an architecture in service of its citizens rather than a Leviathan over them. The Charter is how it writes that architecture down—division by division, what is and what must be—and begins.

This is an analysis published by ENSI (European Nexus for Strategic Intelligence). The current-state baseline reflects real Czech systems and documents—the base registers (ROB, ROS, RÚIAN, RPP), NIA, BankID, eDoklady, datové schránky/ISDS, Czech POINT, the Portál občana, the propojený datový fond (ISZR, eGSB/ISSS), the eGovernment cloud and CLOUDIA, the DIA, Act 12/2020 Sb., the National AI Strategy 2030, Digitální Česko, DESI 2024, and the EU AI Act 2024/1689 (with the CJEU SCHUFA ruling, C-634/21). Figures are reproduced from those sources; the twelve-division architecture and the agentic targets are the author’s coinage.

Sixteen operating principles for a government whose agents navigate the offices on the citizen’s behalf, ask once and never copy, keep a human holding the pen, and ship within today’s law—grounded in Estonia, Ukraine, Singapore, the United Kingdom, and the Czech digital stack.

For twenty years we digitized the state without changing it. We took the form, the queue, and the counter, and we put them on a screen. The portal replaced the window, the PDF replaced the paper, the login replaced the stamp—and the citizen still did all the work. The deep structure never moved: the human being remained the integrator of the state, the living API who carries a birth certificate from the registry to the health insurer to the tax office, who proves the same fact to four agencies that already know it, who assembles the government by hand because the government will not assemble itself. The agentic state is the moment that job ends. These are its sixteen principles.

The first principle is The Citizen Is No Longer the Integrator. This is the spine from which everything else hangs. In the bureaucratic state and in its digital successor, the citizen is the connective tissue between siloed offices—the one who must know which agency needs what, in what order, by when. The agentic state takes that job away from the human and gives it to an agent, and in doing so changes not the interface but the actor.

The second principle is The Life Event Is the Unit of Service. The agentic state is not organized around forms or ministries but around the moments of a human life—a child is born, a job is lost, a business is started, a flood destroys a home. The citizen states the situation in plain language—“help me after the flood”—and the agent orchestrates insurance, housing, and building permits as one act, not as four queues.

The third principle is Ask Once, Never Copy. The agentic state does not hold a thousand duplicate copies of your address. It queries the authoritative register at the moment of decision and never duplicates the data—Estonia’s once-only principle rendered as architecture, where the agent asks the source rather than maintaining yet another stale copy of the truth.

The fourth principle is Predict, Then Offer—Never Impose. The agentic state detects the need before the citizen asks and brings the offer to them: when a family’s income falls below a threshold, the state proposes help with a pre-filled application before they learn it exists. But it offers; it does not compel. Estonia already pays 99.99% of parental benefits automatically; the agentic state generalizes that to every life event, with consent at the gate.

The fifth principle is Any Surface, One Continuous Conversation. Voice, text, phone, computer—start the conversation in the car and finish it at home. The state adapts to the citizen’s channel and context rather than forcing the citizen to adapt to the state’s opening hours and office geography.

The sixth principle is The Official Holds the Pen. The agent prepares; a named human approves every decision that touches a citizen’s rights. This is not only an ethical commitment—it is the precise legal-engineering move that lets the agentic state ship inside the existing administrative code with no statutory change, because the decision-maker of record remains the official, exactly as today.

The seventh principle is Ship Within Today’s Law First. The agentic state does not wait for new legislation. Phase one runs entirely within existing law—it begins with what can be done tomorrow, and legislates afterward, from proven pilots, rather than theorizing rules for a system that does not yet exist.

The eighth principle is Found It by Resolution, Not Statute. The founding act is a government resolution that defines the goal and assigns responsibility—a political mandate and a clear signal, not a new law. The agentic state is launched by executive will and accountability, not by waiting years for a parliament.

The ninth principle is Officials Before Citizens. The first agent serves the civil servant, not the public—lowest risk, highest feedback. An internal assistant on one ministry teaches the state how agents behave before a single citizen’s case depends on one.

The tenth principle is Agents Compose Across Ministries. The architecture is not one monolithic app per office but agents that assemble themselves dynamically across departments to serve a single request. The orchestration layer—the ability to compose a flood-response or a new-business service from many agencies at once—is the moat, not any individual app.

The eleventh principle is The Register Is the Single Source of Truth. Beneath the agents lies the connected data fund: authoritative registers that the agents read, never shadow copies they maintain. The Czech propojený datový fond is the substrate; the agent’s job is to query it, not to recreate it.

The twelfth principle is Build on What We Already Have. The Czech Republic is not a greenfield—it has electronic identity, base registers, and data mailboxes that many countries are only now creating. The problem is not the absence of foundations; it is that we build on them slowly and in fragments. The agentic state is an acceleration and an integration, not a demolition.

The thirteenth principle is Sovereign-European Runtime by Construction. The agentic state runs on inspectable, EU-hostable models, the European identity wallet, and full conformance with the AI Act—so the cognition of the Czech state is auditable, revocable, and sovereign, never rented opaquely from a foreign power that decides what its institutions may say.

The fourteenth principle is Minimization Is the Privacy Firewall. Because the agent asks once and never copies, there is no central super-profile to leak or abuse. Data minimization is not a compliance afterthought bolted onto the system—it is the system’s architecture, the structural reason the agentic state can be proactive without becoming a panopticon.

The fifteenth principle is Contestability with a Named Defendant. Every decision carries an inspectable reason and an affordable appeal to an accountable human—the principle, established in European law by the SCHUFA ruling, that a computation which determines an outcome is itself the regulated decision. There is always an answer, always a defendant, never a faceless machine.

The sixteenth principle is The Manual Fallback Never Dies. A non-digital path always exists, and the state can always be run by hand. The agentic state refuses the monoculture: it preserves the human capacity and the manual procedure so that no failure, no outage, and no excluded citizen is left without recourse.

This article is a field guide to the Agentic State. It states sixteen principles, and dissects each one identically—the Principle itself, its Place in the agentic state across five aspects, the seven reasons it holds, three patterns of how it works in practice, ten building blocks, the four “from → to” shifts it makes, the concrete moves to build it with a real example, and an honest ledger of advantages and risks. It closes with a phased Action Plan anchored to the Czech context and a named deliverable: the Agentic State Operating Charter.

Summary

1) The Citizen Is No Longer the Integrator

• The commitment: the agent, not the human, connects the siloed offices of the state.

• The mechanism: the state assembles itself around a request instead of making the citizen assemble it.

• The payoff: the end of carrying documents between agencies that already hold them.

• The failure it ends: the citizen as unpaid clerk of the bureaucracy.

2) The Life Event Is the Unit of Service

• The commitment: organize service around life moments, not forms or ministries.

• The mechanism: one stated need triggers a coordinated, cross-agency response.

• The payoff: “help me after the flood” instead of four separate queues.

• The failure it ends: the citizen decomposing their own life into agency-shaped tasks.

3) Ask Once, Never Copy

• The commitment: query the authoritative register at decision time; never duplicate.

• The mechanism: the agent asks the source instead of maintaining a stale copy.

• The payoff: the once-only principle made architectural, not aspirational.

• The failure it ends: proving the same fact to agencies that already know it.

4) Predict, Then Offer—Never Impose

• The commitment: detect the need and bring the offer; let the citizen decline.

• The mechanism: threshold-triggered, pre-filled proposals with consent at the gate.

• The payoff: entitlement reaches everyone owed it, without coercion.

• The failure it ends: the take-up gap—help no one knows to claim.

5) Any Surface, One Continuous Conversation

• The commitment: voice, text, any device; resumable across context.

• The mechanism: a single conversation that follows the citizen, not the office hours.

• The payoff: start in the car, finish at home; the state adapts, not you.

• The failure it ends: the citizen bending to the channel and geography of the state.

6) The Official Holds the Pen

• The commitment: agent prepares, a named official approves every rights decision.

• The mechanism: the human decision-maker of record is preserved unchanged.

• The payoff: deployable within the existing administrative code—no statute change.

• The failure it ends: the crumple zone—an unaccountable machine behind a signature.

7) Ship Within Today’s Law First

• The commitment: phase one runs entirely under existing legislation.

• The mechanism: begin with what is legal tomorrow; legislate after pilots prove out.

• The payoff: speed—no waiting years for a new legal regime to start.

• The failure it ends: paralysis-by-regulation that ships nothing.

8) Found It by Resolution, Not Statute

• The commitment: a government resolution defines the goal and assigns responsibility.

• The mechanism: executive mandate and accountability, not a new law.

• The payoff: a clear political signal that can start immediately.

• The failure it ends: the founding hostage to a multi-year legislative cycle.

9) Officials Before Citizens

• The commitment: the first agent serves the civil servant, not the public.

• The mechanism: lowest-risk, highest-feedback internal deployment first.

• The payoff: the state learns agent behavior before citizens’ cases depend on it.

• The failure it ends: a high-stakes public rollout with no operational experience.

10) Agents Compose Across Ministries

• The commitment: agents assemble dynamically across departments for one request.

• The mechanism: an orchestration layer, not a monolithic app per office.

• The payoff: the cross-agency service is the product and the moat.

• The failure it ends: a hundred disconnected ministry apps the citizen must still wire together.

11) The Register Is the Single Source of Truth

• The commitment: agents read authoritative registers, never shadow copies.

• The mechanism: the connected data fund as the substrate of every decision.

• The payoff: one truth, queried—not a thousand drifting duplicates.

• The failure it ends: contradictory records across uncoordinated databases.

12) Build on What We Already Have

• The commitment: exploit existing eID, base registers, and data mailboxes.

• The mechanism: integrate and accelerate the foundations, don’t rebuild them.

• The payoff: a head start most countries lack.

• The failure it ends: greenfield fantasy that ignores real assets and wastes them.

13) Sovereign-European Runtime by Construction

• The commitment: inspectable, EU-hostable models; the eID wallet; AI-Act conformance.

• The mechanism: auditable, revocable cognition the state controls.

• The payoff: the state’s institutions decide what they may conclude and say.

• The failure it ends: governance rented opaquely from a foreign runtime.

14) Minimization Is the Privacy Firewall

• The commitment: ask-once-never-copy means no central super-profile.

• The mechanism: privacy as architecture, not as a compliance bolt-on.

• The payoff: proactive service without a panopticon.

• The failure it ends: the total-surveillance version of proactive government.

15) Contestability with a Named Defendant

• The commitment: every decision carries a reason and an affordable appeal to a human.

• The mechanism: the SCHUFA principle—the determining computation is the decision.

• The payoff: there is always an answer and always a defendant.

• The failure it ends: mass automated harm with no one to confront.

16) The Manual Fallback Never Dies

• The commitment: a non-digital path always exists; the state can run by hand.

• The mechanism: preserved human capacity, manual procedure, model diversity.

• The payoff: no outage, failure, or exclusion leaves a citizen without recourse.

• The failure it ends: the brittle monoculture with no way back.

The Sixteen Principles

1) The Citizen Is No Longer the Integrator

The Principle

The agentic state removes the citizen from the role of integrator—the human being who connects the siloed offices of government by hand—and assigns that role to an agent, so that the state assembles itself around a request instead of demanding that the citizen assemble it.

It functions as the master reframe of the agentic state: the change is not a better interface but a new actor doing the integration work that the human has always, invisibly, done for free.

Place in the Agentic State: 5 aspects

1. The hidden labor of citizenship

   • Every interaction with the bureaucratic state requires the citizen to know which office needs what, in what order.

   • This navigation is unpaid, expert labor the state has always offloaded onto the governed.

2. The digital state did not remove it

   • Twenty years of portals moved the counter to the screen but left the citizen as the connective tissue.

   • A better PDF is still a PDF the citizen must route.

3. The agent as the new connective tissue

   • The agentic state inserts an actor that holds the map of the bureaucracy so the citizen does not have to.

   • The agent knows the agencies, the sequence, and the deadlines.

4. The inversion of burden

   • The work of integration moves from the citizen to the state.

   • The citizen states an intent; the agent executes the orchestration.

5. The spine of every other principle

   • Life-event service, ask-once, proactive offers, cross-ministry composition—all are consequences of this one move.

   • Remove this principle and the rest collapse into “a nicer website.”

Why it holds: 7 reasons

1. The integration work is real — someone must connect the offices; today it is the citizen, at great cost.

2. Agents can hold the whole map — a model can know every agency, rule, and sequence at once.

3. The citizen cannot — no human masters the full topology of the state they must navigate.

4. The data already exists — the offices already hold what the citizen is forced to re-supply.

5. It is the true differentiator — this, not chat, is what makes the state “agentic” rather than “more digital.”

6. It compounds — once the agent integrates, every downstream principle becomes possible.

7. It is humane — it returns time, dignity, and certainty to people at their most vulnerable moments.

Three patterns of how it works

1. Intent → orchestration → result

   • The citizen states an intent in plain language

   • The agent decomposes it into the agencies and steps involved

   • The agent executes and returns the result

2. Map → sequence → execute

   • The agent holds the topology of the bureaucracy

   • It computes the correct sequence of actions

   • It carries them out across offices

3. Silo → bridge → whole

   • Offices remain internally siloed

   • The agent bridges them at the moment of need

   • The citizen experiences one coherent state

Ten building blocks

1. A citizen-facing agent (the new integrator)

2. A map of the bureaucracy (agencies, rules, sequences)

3. Cross-agency orchestration (the composition layer)

4. Authoritative registers (the data the agent queries)

5. Identity and authentication (who the citizen is)

6. Intent understanding (plain-language need → structured action)

7. Task decomposition (need → agency-shaped steps)

8. Status tracking (where each step stands)

9. The official’s approval gate (for rights decisions)

10. An audit trail (what the agent did, on whose behalf)

The shift it makes: four “from → to” moves

1. From citizen-as-clerk to citizen-as-principal

   • The human states intent rather than executing process.

2. From human integration to agent integration

   • The connective work moves from the governed to the state.

3. From interface upgrade to actor change

   • The novelty is a new worker, not a new screen.

4. From navigating the state to being served by it

   • The state assembles itself around the person.

How to build it

A. Put one agent in front of the whole state

• Give the citizen a single conversational entry point that reaches every agency, not one app per office.

• Example: Ukraine’s Diia consolidated services into one experience and helped move the country from 102nd toward the top of the UN e-government ranking; the agentic step adds an agent that orchestrates across them.

B. Hold the map centrally, keep the offices as they are

• Build the orchestration map without forcing every agency to re-platform.

• Example: the Czech propojený datový fond lets an agent query base registers without each office surrendering its system.

C. Measure success as steps removed from the citizen

• Track citizen-initiated inter-agency steps and drive them toward zero.

• Example: the UK’s Tell Us Once lets a death be reported a single time and propagates it to every relevant office—an early, narrow instance of the citizen ceasing to integrate.

Advantages and risks

Advantages

1. The end of the citizen’s hidden, unpaid navigation labor.

2. Time and dignity returned at life’s hardest moments.

3. A genuine generational change, not a cosmetic one.

4. The enabling move for every other principle.

Risks

1. Concentrates enormous orchestration power in one layer—governance is essential.

2. If the agent errs, it errs across many agencies at once.

3. Requires real cross-agency cooperation, which is politically hard.

4. Can mask, rather than fix, broken underlying processes if used as a veneer.

2) The Life Event Is the Unit of Service

The Principle

The agentic state organizes itself around the events of a human life—birth, job loss, starting a business, disaster—rather than around forms, departments, or legal procedures, so that a single stated need triggers a coordinated response across every relevant office.

It functions as the organizing grammar of the agentic state: the citizen describes a situation, not a procedure, and the state translates it into action.

Place in the Agentic State: 5 aspects

1. Life is not shaped like the org chart

   • A flood, a birth, a layoff cuts across many agencies at once.

   • The bureaucratic state forces the citizen to slice their life into agency-shaped pieces.

2. The life event as the interface

   • The citizen states the event—”I had a child,” “I lost my job”—and that is the whole request.

   • The decomposition into tasks is the state’s job, not the citizen’s.

3. Coordination as the deliverable

   • The value is the coordinated bundle: insurance plus housing plus permits, together.

   • A single life event resolves into a single coherent service.

4. The demonstrable difference

   • Life events are where the gap between the digital and the agentic state is instantly visible.

   • They are the proof, not the slogan.

5. The pilotable unit

   • A single life event (a birth) is a clean, bounded first pilot.

   • It generalizes outward to every other event once proven.

Why it holds: 7 reasons

1. It matches reality — people experience events, not procedures.

2. It bundles correctly — one event implies a known set of services that belong together.

3. It is legible — “help after the flood” is a request anyone can state.

4. It is demonstrable — the contrast with today is immediate and visceral.

5. It is bounded — a single event is a tractable pilot scope.

6. It compounds — events share components (identity, registers, approval) reusable across all.

7. It is humane — it meets people in the language of their lives, not the state’s.

Three patterns of how it works

1. Event → bundle → delivery

   • A life event is declared

   • The state maps it to its bundle of services

   • The bundle is delivered as one

2. Trigger → orchestration → completion

   • A trigger (birth registered, income dropped) fires

   • The agent orchestrates the relevant agencies

   • The service completes with minimal citizen input

3. One event → many agencies → one experience

   • A single event touches many offices

   • The agent coordinates them

   • The citizen sees one seamless response

Ten building blocks

1. A catalog of life events (the service grammar)

2. Event-to-service mappings (what each event implies)

3. Triggers (registrations and thresholds that fire events)

4. Cross-agency orchestration (the bundle delivery)

5. Pre-filled applications (from authoritative data)

6. Consent capture (the citizen’s go-ahead)

7. The official’s approval (for rights decisions in the bundle)

8. Status and notification (keeping the citizen informed)

9. Reusable components (identity, registers, approval, shared across events)

10. Outcome tracking (did the bundle actually help)

The shift it makes: four “from → to” moves

1. From form to life event

   • The unit of service becomes the human moment, not the document.

2. From citizen-side decomposition to state-side decomposition

   • The state slices the event into tasks, not the citizen.

3. From sequential queues to one coordinated bundle

   • Services arrive together, not one office at a time.

4. From procedure-first to need-first

   • The citizen states a need; procedure becomes the state’s internal concern.

How to build it

A. Start with one high-clarity event

• Pick a single, emotionally clear life event and deliver it end-to-end.

• Example: the birth of a child—the maternity ward registers the birth, and the state proactively offers everything the family needs in a few clicks—is the canonical first pilot.

B. Map the bundle before building the agent

• For each event, enumerate the agencies and services that belong together.

• Example: Singapore’s LifeSG organizes government around life moments and cross-agency data sharing, bundling services a citizen would otherwise chase separately.

C. Generalize by reusing components

• Build identity, register-query, and approval once; reuse them across every event.

• Example: once “birth” works, “job loss” and “starting a business” reuse the same orchestration spine.

Advantages and risks

Advantages

1. Service that matches how people actually live.

2. Immediate, demonstrable improvement over the digital state.

3. Clean, bounded pilots that generalize.

4. Reusable components that compound across events.

Risks

1. Event bundles can embed wrong assumptions about what people need.

2. Edge cases and unusual life situations may be poorly served.

3. Bundling can over-reach, offering more than the citizen wants.

4. Requires cross-agency agreement on what each event entails.

3) Ask Once, Never Copy

The Principle

The agentic state collects a fact from the citizen at most once and thereafter queries the authoritative register at the moment of decision—never duplicating, syncing, or hoarding a private copy of data the state already holds.

It functions as the data constitution of the agentic state: the once-only principle rendered as architecture rather than aspiration.

Place in the Agentic State: 5 aspects

1. The citizen is not a data courier

   • In the bureaucratic state, the citizen re-supplies the same facts to every office.

   • Ask-once ends the courier role: the state already holds the truth.

2. Query, do not copy

   • The agent reads the source register when it needs a fact, then forgets it.

   • No new master copy is created to drift, leak, or contradict.

3. The register as truth

   • The authoritative register is the single source; everything else queries it.

   • Copies are the enemy of consistency and of privacy alike.

4. Privacy by minimization

   • Because nothing is copied, there is no central super-profile to abuse.

   • The data constitution is also the privacy firewall.

5. Consent and purpose at the query

   • Each query is purpose-bound and, where required, consented.

   • Access is logged at the point of use, not buried in a warehouse.

Why it holds: 7 reasons

1. The state already has the data — re-collection is pure waste and indignity.

2. Copies drift — duplicated data becomes inconsistent and wrong.

3. Copies leak — every duplicate is a new attack surface.

4. Querying is now cheap — connected registers make real-time lookup feasible.

5. It minimizes by design — no hoard means less to protect and abuse.

6. It localizes truth — one authoritative source ends contradictory records.

7. It is auditable — purpose-bound queries log who saw what, when, and why.

Three patterns of how it works

1. Need → query → forget

   • A decision needs a fact

   • The agent queries the authoritative register

   • The fact is used and not retained

2. Source → authority → consistency

   • One register is authoritative for a fact

   • All consumers query it

   • The whole state stays consistent

3. Consent → purpose-bound access → log

   • The citizen consents to a use

   • Access is bound to that purpose

   • The access is logged for audit

4. (patterns held to three)

Ten building blocks

1. Authoritative base registers (the sources of truth)

2. A connected data fund (the query fabric)

3. Real-time query interfaces (lookup at decision time)

4. No-copy data policies (prohibition on duplication)

5. Purpose-binding (each access tied to a reason)

6. Consent management (where consent is required)

7. Access logging (who queried what, when)

8. Data-quality governance (the source must be correct)

9. Identity resolution (linking citizen to record)

10. Selective disclosure (reveal the minimum needed)

The shift it makes: four “from → to” moves

1. From re-supply to ask-once

   • The citizen provides a fact at most once.

2. From copy to query

   • The state reads the source instead of hoarding duplicates.

3. From data warehouse to data fund

   • Truth lives in authoritative registers, queried on demand.

4. From bolt-on privacy to architectural privacy

   • Minimization is built into how data is accessed, not added later.

How to build it

A. Forbid the copy

• Make “query the register, do not duplicate” a hard architectural rule for every agent.

• Example: Estonia’s once-only principle and X-Road data exchange let agencies ask the source rather than maintain copies—the model the Czech propojený datový fond extends.

B. Make every query purpose-bound and logged

• Bind each data access to a stated purpose and record it for the citizen to inspect.

• Example: a citizen-visible access log so anyone can see which office queried which fact, and why.

C. Invest in the authority of the source

• A query architecture is only as good as the register it queries—fund data quality.

• Example: designate and maintain base registers as the legally authoritative source for each class of fact.

Advantages and risks

Advantages

1. The end of re-supplying facts the state already holds.

2. Consistency—one truth, queried, not many copies drifting.

3. Privacy by minimization—no central hoard to abuse.

4. Auditability—purpose-bound, logged access.

Risks

1. A wrong fact in the authoritative register propagates everywhere.

2. Real-time query availability becomes mission-critical infrastructure.

3. Centralized query fabric is itself a high-value target.

4. Purpose-binding must be enforced, not merely declared.

4) Predict, Then Offer—Never Impose

The Principle

The agentic state detects a citizen’s need before they ask and brings them a ready-made offer—a pre-filled application, a calculated benefit—while leaving the decision to accept entirely with the citizen.

It functions as the proactivity clause of the agentic state: anticipation with consent, help that arrives early but never imposes itself.

Place in the Agentic State: 5 aspects

1. The take-up gap is a failure

   • Benefits that people are entitled to but never claim are rights in name only.

   • Proactivity closes the gap by bringing the offer to the person.

2. The state acts first

   • When income drops below a threshold, the state offers help before the family knows it exists.

   • The initiative shifts from citizen to state.

3. An offer, not an order

   • The state proposes; the citizen disposes. Acceptance is always the citizen’s.

   • Proactivity without consent is coercion; the line is absolute.

4. Pre-filled, not pre-decided

   • The application arrives complete, but the citizen confirms and submits.

   • The work is done; the choice remains.

5. Timed to the moment

   • The offer arrives when it is useful—at the birth, at the layoff, after the flood.

   • Relevance is a function of timing, and the agent gets the timing right.

Why it holds: 7 reasons

1. Entitlement should reach its target — unclaimed help is a policy failure, not the citizen’s fault.

2. Friction is regressive — application burdens fall hardest on those most in need.

3. The data enables it — thresholds and triggers are computable from registers the state holds.

4. Consent preserves autonomy — offering, not imposing, keeps the citizen sovereign.

5. Timing multiplies value — help at the right moment is worth far more than help eventually.

6. It builds trust — a state that anticipates and offers earns legitimacy.

7. It is already proven — Estonia pays 99.99% of parental benefits automatically.

Three patterns of how it works

1. Trigger → calculate → offer

   • A threshold or event fires

   • The agent calculates the entitlement

   • It offers a pre-filled application

2. Detect → propose → consent

   • The need is detected

   • The state proposes a remedy

   • The citizen consents or declines

3. Profile → match → time

   • The citizen’s situation is understood

   • It is matched to relevant support

   • The offer is timed to the moment of need

Ten building blocks

1. Eligibility triggers (thresholds and events)

2. Entitlement calculators (compute the benefit)

3. Pre-filled applications (from authoritative data)

4. Consent gates (acceptance is the citizen’s)

5. Timing logic (deliver at the useful moment)

6. Profile matching (situation → relevant support)

7. Notification channels (reach the citizen)

8. Decline and opt-out paths (refusal is easy)

9. The official’s approval (for the rights decision)

10. Outcome monitoring (did the offer help)

The shift it makes: four “from → to” moves

1. From apply to be offered

   • The state brings the offer; the citizen need not chase it.

2. From reactive to proactive

   • Help arrives before the request, not after.

3. From paperwork to confirmation

   • The application is pre-filled; the citizen confirms.

4. From eventual to timely

   • Support lands at the moment it matters.

How to build it

A. Compute the trigger from data you already hold

• Define thresholds and events that the connected registers can detect automatically.

• Example: when a family’s income falls below a defined line, the state offers help with a pre-filled application before they learn of it.

B. Make declining trivial

• Every proactive offer must be as easy to refuse as to accept.

• Example: a single “no thanks” that the agent respects and logs—proactivity that never becomes pressure.

C. Generalize from a proven automatic service

• Start where automation is already accepted and extend the pattern.

• Example: Estonia’s near-fully-automatic parental benefit (99.99%) is the template for proactive offers across every life event.

Advantages and risks

Advantages

1. The take-up gap closes—entitlement reaches everyone owed it.

2. Help arrives at the moment it is most useful.

3. Friction and indignity are removed for the most vulnerable.

4. Trust grows in a state that anticipates and offers.

Risks

1. Proactivity without strict consent slides into coercion.

2. Detecting need requires data use that must be tightly minimized.

3. Wrong triggers offer the wrong help to the wrong people.

4. “Helpful” anticipation can feel like surveillance if not transparent.

5) Any Surface, One Continuous Conversation

The Principle

The agentic state meets the citizen on any channel—voice, text, phone, computer—and treats every interaction as one continuous conversation that follows the person across devices and time, rather than a series of disconnected sessions on the state’s terms.

It functions as the access clause of the agentic state: the state adapts to the citizen’s surface and context, not the reverse.

Place in the Agentic State: 5 aspects

1. The channel is the citizen’s choice

   • Voice in the car, text on the phone, a screen at home—whichever suits the moment.

   • The state is present wherever the citizen is.

2. One conversation, not many sessions

   • A request begun on one device continues on another without restarting.

   • Context persists; the citizen does not repeat themselves.

3. Plain language, not forms

   • The citizen speaks or types naturally; the agent translates to action.

   • The interface is conversation, not a field layout.

4. Accessibility by default

   • Voice and natural language open the state to those excluded by complex portals.

   • The least digitally fluent are first-class users.

5. The state adapts, not the citizen

   • Office hours, geography, and channel constraints dissolve.

   • The burden of adaptation moves to the state.

Why it holds: 7 reasons

1. People live across channels — a single mandated interface fits no one perfectly.

2. Context is precious — forcing a restart wastes the citizen’s effort and patience.

3. Natural language is universal — conversation is the most accessible interface there is.

4. It includes the excluded — voice reaches those whom portals leave behind.

5. It is now feasible — agents can sustain context across channels and time.

6. It dissolves friction — no hours, no geography, no channel lock-in.

7. It dignifies — the state coming to the citizen’s surface is respect made operational.

Three patterns of how it works

1. Begin → persist → resume

   • A conversation begins on one surface

   • Context is persisted

   • It resumes seamlessly on another

2. Speak → understand → act

   • The citizen states a need in natural language

   • The agent understands intent

   • It acts across the state

3. Any channel → one identity → one thread

   • The citizen arrives on any channel

   • Authenticated to one identity

   • Continuing one coherent thread

Ten building blocks

1. Multi-channel front ends (voice, text, web, mobile)

2. Persistent conversation state (context that follows)

3. Natural-language understanding (intent from speech/text)

4. Cross-device identity (one authenticated thread)

5. Accessibility features (voice, plain language, assistance)

6. Session continuity (resume, never restart)

7. Channel-appropriate rendering (fit the surface)

8. Offline and low-bandwidth fallback

9. Privacy across channels (consistent protection)

10. Human handoff (to an official when needed)

The shift it makes: four “from → to” moves

1. From the state’s channel to the citizen’s

   • Access happens on the citizen’s chosen surface.

2. From sessions to one conversation

   • Context follows the person across devices and time.

3. From forms to natural language

   • The interface is speech and text, not field layouts.

4. From office hours to always-available

   • Geography and opening times dissolve.

How to build it

A. Persist context across every surface

• Treat identity and conversation state as continuous, never per-device.

• Example: begin a request by voice in the car and complete it on a laptop at home without re-explaining anything.

B. Lead with voice and plain language

• Make natural conversation the primary interface, not an add-on to forms.

• Example: a citizen says “I lost my job” and the agent proceeds—no menu tree, no form codes.

C. Guarantee a human handoff

• Every conversation can escalate to an official when the citizen wants one.

• Example: “I’d rather speak to a person” routes to a named official with the full context attached.

Advantages and risks

Advantages

1. Access on the citizen’s terms, channel, and time.

2. Inclusion of those excluded by complex portals.

3. No repetition—context follows the person.

4. The dignity of a state that comes to you.

Risks

1. Cross-channel context is a privacy and security challenge.

2. Voice and natural language introduce recognition errors.

3. Continuity across devices widens the authentication attack surface.

4. Conversational interfaces can obscure what the state is actually doing.

6) The Official Holds the Pen

The Principle

For every decision that touches a citizen’s rights, the agent prepares the case and a named human official approves it—so the decision-maker of record remains the official, exactly as today, and the agent never decides alone.

It functions as the legal keystone of the agentic state: the precise move that preserves due process and lets the system run inside the existing administrative code without statutory change.

Place in the Agentic State: 5 aspects

1. Preparation, not decision

   • The agent assembles the case, checks conditions, and drafts the determination.

   • The act of deciding remains the official’s.

2. The administrative code is preserved

   • Because the official decides, the existing správní řád applies unchanged.

   • No new legal regime is required to begin.

3. Accountability has a name

   • Every rights decision has a human owner who can be identified and held responsible.

   • There is no “the model decided.”

4. The crumple zone is refused

   • The official is empowered and informed, not a powerless signature on an opaque output.

   • Approval is substantive, not performative.

5. The boundary is bright

   • Routine, non-rights actions the agent may complete; rights decisions it may only prepare.

   • The line between preparation and decision is explicit.

Why it holds: 7 reasons

1. It preserves due process — a human decision-maker is what the law and legitimacy require.

2. It avoids legal change — keeping the official as decider means no statute must move first.

3. It anchors accountability — a named human owns each outcome.

4. It refuses the crumple zone — the official decides, not absorbs blame for a machine.

5. It builds trust — citizens accept being judged by a person assisted by a tool.

6. It is auditable — preparation and approval are distinct, logged steps.

7. It is pragmatic — it lets the agentic state start now, within today’s institutions.

Three patterns of how it works

1. Prepare → present → approve

   • The agent prepares the case

   • It presents the draft to the official

   • The official approves, amends, or rejects

2. Routine → autonomous; rights → human

   • Routine actions complete autonomously

   • Rights decisions route to a human

   • The boundary governs which path applies

3. Draft → review → own

   • The agent drafts the determination

   • The official substantively reviews it

   • The official owns the decision

Ten building blocks

1. A preparation engine (the agent’s casework)

2. The approval interface (where the official decides)

3. The rights/routine boundary (what needs a human)

4. Reason traces (so the official can review meaningfully)

5. Amendment capacity (the official can change the draft)

6. Named accountability (the human owner of record)

7. Audit logs (preparation and approval as distinct events)

8. Override metrics (proof the official actually decides)

9. Escalation paths (complex cases to senior officials)

10. Training (officials as supervisors of agent casework)

The shift it makes: four “from → to” moves

1. From agent-decides to agent-prepares

   • The machine readies the case; the human decides.

2. From new law to existing law

   • Preserving the official as decider keeps the administrative code intact.

3. From crumple zone to empowered approver

   • The official is informed and able to change the outcome.

4. From diffuse blame to named accountability

   • Each decision has a human owner.

How to build it

A. Draw the rights/routine boundary explicitly

• Define which actions the agent may complete and which it may only prepare.

• Example: an agent may file a notification autonomously but may only prepare a benefit determination for an official to approve.

B. Make approval substantive

• Give the official the reasons, the power to amend, and the time to use them; measure overrides.

• Example: the principle stated plainly on the Czech concept—”AI připravuje podklady, úředník schvaluje. Proto není potřeba měnit správní řád.”

C. Keep preparation and approval as separate, logged acts

• Record the agent’s preparation and the official’s decision as distinct, auditable events.

• Example: an audit trail that shows what the agent drafted and what the official decided, separately.

Advantages and risks

Advantages

1. Due process preserved in substance.

2. Deployable within existing law—no statute change needed.

3. A named, accountable human for every rights decision.

4. The crumple zone refused by design.

Risks

1. Automation bias can hollow approval into rubber-stamping—must be measured.

2. Volume can pressure officials toward perfunctory review.

3. The rights/routine boundary will be contested at the edges.

4. Without real override capacity, “approval” becomes theater.

7) Ship Within Today’s Law First

The Principle

The agentic state begins entirely within existing legislation—deploying everything that is already legal—and legislates afterward, from the evidence of working pilots, rather than waiting for a new legal regime before it starts.

It functions as the tempo clause of the agentic state: start with what is possible tomorrow, and let law follow proof.

Place in the Agentic State: 5 aspects

1. Most of it is already legal

   • Preparing cases, querying registers with consent, official approval—all permissible today.

   • The agentic state’s first phase needs no new statute.

2. Law follows evidence

   • New legislation, when needed, is written from working pilots, not from speculation.

   • Proof precedes regulation.

3. Speed as a strategy

   • Beginning now, within the law, captures years that legislating-first would waste.

   • Tempo is itself an advantage.

4. Reform from a position of knowledge

   • Once pilots run, the state knows exactly which laws to change and why.

   • Legal reform becomes targeted, not theoretical.

5. Risk contained by scope

   • Operating within existing law keeps early deployments bounded and reversible.

   • The legal envelope is a safety rail.

Why it holds: 7 reasons

1. The legal room exists — preparation-plus-approval fits the current administrative code.

2. Legislating-first is slow — waiting for new law forfeits years.

3. Pilots teach — running systems reveal which laws actually need changing.

4. Evidence beats theory — laws written from proof are better laws.

5. Speed compounds — early starts accumulate learning and legitimacy.

6. Scope limits risk — the legal envelope bounds early deployments.

7. It is politically achievable — starting needs no parliamentary majority, only executive will.

Three patterns of how it works

1. Within-law pilot → evidence → targeted reform

   • A pilot runs within existing law

   • It produces evidence

   • Targeted legal reform follows

2. Possible-now → deploy → learn

   • Identify what is already legal

   • Deploy it

   • Learn what to change next

3. Bounded scope → expand → legislate

   • Start within a bounded legal envelope

   • Expand as proof accumulates

   • Legislate to enable the next stage

4. (patterns held to three)

Ten building blocks

1. A legal-feasibility map (what is already permissible)

2. Within-law pilot designs (no statute required)

3. The preparation-plus-approval pattern (the legal anchor)

4. Consent and purpose-binding (lawful data use today)

5. Evidence collection (to justify later reform)

6. A reform backlog (laws to change, prioritized by proof)

7. Scope boundaries (the legal envelope)

8. Reversibility (pilots that can be undone)

9. Regulatory liaison (to prepare targeted change)

10. Public transparency (what is being piloted, and why)