ENSI · REPORT · 2026

National Agentic Technology Infrastructure

Architecting the Agent-Native State — how the world's leading states are extending the digital-government canon to a workforce that reasons and acts.

Fewer than ~10 nations have wired the agentic foundation onto a healthy digital state — and perhaps two or three have done it well. That gap is the story of the agentic decade.

Evidence base 174 sources Defining number ≈10 nations Published June 2026 European Nexus for Strategic Intelligence · Issue 01

About this report · How to read it

The agentic state, evidenced


The European Nexus for Strategic Intelligence (ENSI) builds decision-grade intelligence on how nations architect technology. This issue maps the next layer of the state: the public infrastructure that lets AI agents — software that perceives, reasons and acts within bounded authority — do useful, safe, accountable work at population scale. It is not an AI-strategy PDF and not a chatbot procurement; it is a reading of the agentic literature against the national-architecture canon, resolved into a buildable stack.

The evidence base

This report is grounded in a verified collection of 142 downloaded PDFs and ~174 catalogued sources spanning 12 folders — from national architectures (Estonia, Singapore, India, the UAE) and the digital-state frameworks (eIDAS, NIST, FedRAMP) to the agentic corpus: the Agentic State vision paper, NASCIO 2026, OpenAI's governing-practices, the MCP and A2A protocols, and the agent-systems research (ReAct, RAG, Reflexion, MemGPT, τ-bench). Every number traces to a source; agentic adoption figures the literature could not count precisely are labelled (est.).

Three reading depths
  • 90 seconds — this front matter: cover, summary, KPI dashboard, the one-diagram thesis and twelve findings.
  • 15 minutes — add the framework, the cross-country landscape and the ranked architectures.
  • Full read — the 50 principles, 50 components and ten archetypes, with the cases, evidence and appendix.

The visual system — a legend

Magenta is the one accent: emphasis and the load-bearing layers. Green marks data-positive, coral marks an alert. Two coding systems run through every page — the five priority tiers and the eight stack layers.

Priority tiers — how essential a component is

P0 Bedrock — no agentic state without it P1 Core — defines a serious agentic state P2 Differentiator — separates the leaders P3 Advanced — the maturing edge P4 Frontier — the experimental frontier

The eight stack layers — where a component lives

L1 Compute & Models L2 Data, Knowledge & Memory L3 Tools & Integration L4 Trust & Control · cross-cutting L5 Agent Runtime L6 Orchestration L7 Experience & Mission L8 Governance Spine · cross-cutting

Source chips (e.g. NASCIO 2026) close findings and tables; bold renders magenta throughout and always marks the thing that matters on the page.

NATI · ENSI Report 2026The 90-Second Read
ENSI · National Agentic Technology Infrastructure02

Contents

What is inside


Movement A — The 90-second read
01Cover1
02About ENSI · how to read · the visual legend2
03Contents3
04Executive summary — the spine in six paragraphs4
05The dashboard — twelve headline indicators6
06The thesis in one diagram — the 8-layer stack8
07Twelve key findings9
Movement B — Framework & landscape
08The reframe — why "buy an AI" is the wrong question11
09From the digital state to the agentic state13
10The 8-layer National Agentic Technology Stack, explained15
11The global landscape & readiness18
12How we rank — priority tiers & the scoring lens20
Movement C — Principles, components & architectures
13The 50 principles of the agent-native state21
14The 50 components — a bill of materials30
15The adoption curve — where nations fall off37
16The 10 best agentic architecture types, ranked39
17The optimal national agentic stack — the reference design46
Movement D — Evidence & cases
18Who's closest — the agentic-state scorecard49
19Seven nation case studies50
20Cross-cutting — sequence, failure modes & recommendations57
21Risks, counter-arguments & outlook to 203059
Movement E — Back matter
22Methodology & integrity notes61
23The 174-source evidence base — annotated catalogue62
24Back cover — the order is the strategy64

Three depths: 90 seconds — pp. 1–10 · 15 minutes — add the framework, adoption curve & optimal stack · Full read — the 50 principles, 50 components, 10 archetypes & 7 cases. Every component carries a priority (P0P4) and an adoption figure; agentic estimates are labelled (est.).

NATI · ENSI Report 2026The 90-Second Read
ENSI · National Agentic Technology Infrastructure03

Executive summary · 1 of 2

The agentic state is the digital state, extended


For thirty years the organising metaphor of national technology was the digital service — a form turned into a web page, a registry turned into an API. The best states built that layer well: Estonia's X-Road, India's identity-payments-data stack, Singapore's government-as-a-platform, the EU's interoperability frameworks. That achievement is now the floor, not the ceiling. A new substrate has arrived — AI agents — and NASCIO frames the shift bluntly as the move from "AI that assists" to "AI that acts."

National Agentic Technology Infrastructure (NATI) is the public infrastructure a nation must build so that agents can act usefully, safely and accountably at population scale. It is a stack — compute, models, data, memory, tools, identity, orchestration, governance — re-conceived around a worker that reasons and acts. The central claim of this report is that the digital state and the agentic state are not rivals: the agentic state is what you build on top of a working digital state, and it fails without one. An agent is only as good as the identity it can verify, the registry it can trust, the tool it can call, and the human who can stop it.

The optimal design at this stage is clear and convergent. It is a sovereign agent platform provisioned centrally for all of government, on which every agent is a grounded, reflective reason–act–verify loop with governed memory, reaching the world through an MCP tool layer and other agents through an A2A federation, all running under supervised autonomy — bounded action, mandatory human gates, agent identity, authenticated delegation, full observability, a tested kill-switch — evaluated continuously against a mandatory eval harness and a binding national risk framework, and sitting atop a healthy digital state.

The win is not any single exotic pattern. It is having all of the agentic P0–P1 components wired together onto a working digital foundation — which fewer than ~10 nations have done, and perhaps two or three have done well. The decisive strategic fact is that the agentic race is being run on the foundations laid by the digital race: nations that built X-Road-class interoperability, population-scale identity and clean registries already hold half the agentic P0 stack, while those that skipped the digital foundation will find that no amount of agent procurement substitutes for it.

The spine — the claim this whole report defends

The agentic state is the digital state's canon — identity, registries, federation, interoperability, governance, oversight — extended to a worker that reasons and acts. The win is having the agentic P0–P1 layer wired onto a healthy digital foundation, which fewer than ~10 nations have done.

The pages that follow translate this into a buildable architecture: a priority model from P0 Bedrock to P4 Frontier, an eight-layer stack, fifty principles, fifty components and ten archetypes — and a scorecard of who, today, is closest to the optimal design.

NATI · ENSI Report 2026The 90-Second Read
ENSI · National Agentic Technology Infrastructure04

Executive summary · 2 of 2

What separates a thinking state from a demo


Two paragraphs of consequence, then the five things a decision-maker must internalise.

The defining risk of agents is confident error; the defining requirement of public administration is defensible decisions. Grounding resolves the tension — retrieval from authoritative registries, with citations, turns an agent's output into something it can prove, so ungrounded agents on official matters automate unaccountability. And because models are replaced every few months, a continuous evaluation harness measuring capability and consistency is the difference between a controlled rollout and a silent regression — τ-bench's finding that agents often fail on repeated trials is the single most important deployment caution in the evidence base.

The other half of the answer is governance as architecture. For software that acts, identity, authenticated delegation, least-privilege scopes, human approval gates and a tested kill-switch are not a brake bolted on afterwards — they are structural members of the design. Supervised autonomy is the only pattern present in every credible governance source, and the maturity discipline — climb, do not leap — is how a state earns autonomy as reliability is proven. The order is the strategy.

Five things to know
  1. It's continuity, not rupture. The agentic state inherits the digital canon and extends it to a worker that acts — it does not replace it.
  2. The gap is the story. Fewer than ~10 nations have the agentic P0–P1 layer wired onto a healthy digital foundation; two or three have done it well.
  3. Build a platform, not pilots. One shared sovereign platform operationalises safety once and inherits it everywhere; a thousand pilots cannot be governed.
  4. Earn the AI layer. Intelligence is the capstone of identity → data → interoperability — resource the lower layers first or automate dysfunction.
  5. Supervised autonomy is the architecture. Bounded action, human gates, agent identity and a kill-switch make autonomous action legally survivable.

No nation has the full optimal stack. The frontier is a composite — the US's platform, Estonia's digital foundation and framework, the UK's assurance discipline, Singapore's governance toolkit, the EU's binding risk law, and the UAE's mission ambition. The report's task is to name the components, rank the architectures, and show the assembly order.

~10nations with the agentic P0–P1 layer on a healthy digital state (est.)
8+US states already running agentic tools in production — NASCIO 2026
50design principles · 50 components · 10 architecture archetypes
NATI · ENSI Report 2026The 90-Second Read
ENSI · National Agentic Technology Infrastructure05

The dashboard · 1 of 2

The agentic state in twelve numbers


The foundation and the frontier, read together. The connectivity, identity and cyber numbers are hard global figures; the agentic-adoption tiles are labelled (est.) where the literature could not count precisely.

5.5bn
people online in 2024 — 68% of humanity; 2.6bn remain offline
ITU FACTS & FIGURES 2024
96%
of the world covered by a mobile-broadband network
ITU 2024
84/4%
5G population coverage — high-income vs low-income countries
ITU 2024
139/194
nations with a national CIRT / CERT; 55 still have none
ITU GCI 2024
~850m
people still lack any official ID — the agent's root of trust
WORLD BANK ID4D 2021
8+
US states already running agentic tools in production
NASCIO 2026
~10
nations with the full agentic P0–P1 stack on a healthy digital state (est.)
ENSI SYNTHESIS — EST.
~5–15
nations standing up a national agent platform (est.)
ENSI SYNTHESIS — EST.
~38
nations benchmarked on digital-government maturity
OECD DGI 2023
~20–25
nations with governed model-access programmes (est.)
ENSI SYNTHESIS — EST.
Reading the dashboard

The top row is the digital foundation the agentic state is built on — broadband, coverage, 5G, cyber response. The lower rows are the agentic frontier, where hard counts thin out and the synthesis estimates begin: the population of nations with a wired P0–P1 stack is small, and the platform builders are fewer still. Estimates are derived from the leaders documented in folders 08–12 and are explicitly indicative.

NATI · ENSI Report 2026The 90-Second Read
ENSI · National Agentic Technology Infrastructure06

The dashboard · 2 of 2

Who leads — and the adoption cliff


Two hero charts and a share view. The leaderboard is an indicative composite against the optimal stack (0–100, est.); the drop-off shows how adoption collapses as you climb the priority tiers.

FIG 1 · indicative / est.
United States (federal)82
Estonia76
Singapore74
United Kingdom71
EU (as a bloc)68
UAE64
Leading US states61
Who is closest to the optimal agentic stack. Indicative composite score (0–100, est.) blending platform, grounding, supervised-autonomy, identity/delegation and digital-foundation strength. No nation has the full stack — the leaders are complementary strengths, not a single winner. Source: ENSI synthesis of folders 08–12.
FIG 2 · est.
P0 Bedrock~many
P1 Core~10
P2 Differ.~6
P3 Advanced~3
P4 Frontier~2
The adoption cliff. Many states hold the P0 digital bedrock; the count of nations falls steeply as the priority tier rises, with only ~2 reaching the frontier (est.). The gap between P0 and P1 is the agentic decade's central opportunity.
FIG 3 · est.
~5%
Nations with a wired agentic P0–P1 stack
All other UN member states (~183)
The ~10-nation share. Roughly 10 of 193 UN member states have wired the agentic foundation onto a healthy digital state — about 5% (est.). The agentic state is, today, a minority achievement.
Caveat — read the estimates as indicative

No global index yet counts "agentic-ready" nations. FIG 1–3 are ENSI composites built from the documented leaders (USAi, the Agentic State framework, NASCIO's production states, NAIS 2.0, the EU AI Act) and are directional, not measured.

NATI · ENSI Report 2026The 90-Second Read
ENSI · National Agentic Technology Infrastructure07

The master diagram

The 8-layer National Agentic Technology Stack


Read top to bottom. Magenta bands are the load-bearing layers; each carries its name, one-line role, and its P0 component. Everything sits on the digital-state foundation — remove it and the stack falls.

FIG 4 · the reference architecture
L8
Governance Spine · cross-cutting — binding risk framework, liability law, central agent authority. P0: national risk-management spine
L7
Experience & Mission — citizen super-assistants, civil-servant copilots, policy twins — bounded and human-gated. P0: human-gated mission services
L6
Orchestration — durable workflows, orchestrator-worker multi-agent, A2A inter-agency federation, human-approval nodes. P1: stateful orchestration runtime
L5
Agent Runtime — the reason–act–verify loop, reflection & self-correction, planning, skill library. P0: grounded reason–act–verify loop
L4
Trust & Control · cross-cutting — agent identity, authenticated delegation, least-privilege scopes, guardrails, kill-switch, observability. P0: agent identity + authenticated delegation
L3
Tools & Integration — MCP servers over every registry/system, tool & agent discovery, sandboxed legacy bridge. P0: MCP tool layer + secure gateways
L2
Data, Knowledge & Memory — agent-ready registries, RAG, vector + graph stores, governed tiered memory, consent layer. P0: agent-ready registries + RAG grounding
L1
Compute & Models — governed frontier + sovereign/open-weight models, a model gateway, managed national inference. P0: governed model access + gateway
▼ Sits atop the ▼
DIGITAL STATE FOUNDATION
universal broadband · eID · base registries · X-Road-style data exchange · sovereign cloud · zero-trust security · interoperability — the agentic stack fails without it
The optimal National Agentic Technology Stack (2026). The agentic state is the digital state's canon — identity, registries, federation, governance, oversight — extended to a worker that reasons and acts; the win is the agentic P0–P1 layer wired onto a healthy digital foundation, which fewer than ~10 nations have done. Source: ENSI synthesis document 3 of 3.
NATI · ENSI Report 2026The 90-Second Read
ENSI · National Agentic Technology Infrastructure08

Key findings · 1 of 2

Twelve findings — the first six


The load-bearing conclusions of the evidence base, each closing with its strongest source. Findings 1–6 cover the strategy and the architecture of trust.

  1. It is continuity, not rupture. The agentic state inherits the digital state's canon — identity, registries, federation, interoperability, governance — and extends it to a worker that reasons and acts; it does not replace it. Algorithmic State Architecture, arXiv 2503.08725
  2. Fewer than ~10 nations have closed the gap. Only a handful have wired the agentic P0–P1 layer onto a healthy digital state, and perhaps two or three have done it well — the central strategic fact of the agentic decade. ENSI synthesis (est.)
  3. Build a platform, not pilots. The leaders provision a shared sovereign agent platform — model access, guardrails, observability and a tool catalogue — so every ministry inherits safety instead of re-failing at it. GSA USAi · EU GenAI4EU
  4. Grounding is the antidote to confident error. Every consequential agent must retrieve from authoritative sources and cite them; ungrounded generation on official matters is a defect, not a feature. RAG, arXiv 2005.11401 · OECD/GPAI
  5. Supervised autonomy is the architecture. Bounded action spaces, mandatory human gates, escalation paths and a tested kill-switch are the only pattern present in every credible governance source — a structural member, not a brake. EU AI Act · NASCIO 2026 · UK AI Playbook
  6. Every agent needs a verifiable identity and scoped delegation. An action you cannot attribute is an action you cannot govern; agents act only under least-privilege, time-bounded, auditable, revocable delegation. Visibility into AI Agents, arXiv 2401.13138 · MIT, arXiv 2501.09674
The throughline

Findings 1–6 establish the strategy (continuity, the gap, the platform) and the architecture of trust (grounding, supervised autonomy, agent identity). Findings 7–12, overleaf, cover how agents reach the world, how they are proven reliable, and where the adoption curve breaks.

NATI · ENSI Report 2026The 90-Second Read
ENSI · National Agentic Technology Infrastructure09

Key findings · 2 of 2

Twelve findings — the second six


Findings 7–12 cover integration, reliability, data readiness, the leaders, the cliff, and the rule that governs the whole sequence.

  1. MCP + A2A are federation reborn at the agent layer. The Model Context Protocol (agent-to-tool) and Agent2Agent (agent-to-agent) reproduce X-Road's federation for agents — integrate through open standards, preserve agency ownership, avoid a central black box. Anthropic MCP · A2A → Linux Foundation
  2. No deployment without an eval harness — and continuous re-evaluation. Agents must be measured for capability and consistency; τ-bench's pass^k shows they often fail on repeated trials, and every model change demands re-qualification. τ-bench · AgentBench · GAIA
  3. Data readiness is a precondition, not a parallel workstream. An agent automates whatever quality it is fed, at scale; agent-ready, clean, machine-readable registries are the new precondition the way machine-readable data once was. OECD/GPAI government-data report
  4. No nation has the full stack — the frontier is a composite. The US's platform, Estonia's foundation and framework, the UK's assurance, Singapore's governance toolkit, the EU's risk law and the UAE's mission ambition. ENSI scorecard, folders 08–12
  5. The adoption curve falls off a cliff above P0. Many states hold the P0 digital bedrock; the count collapses through P1–P4, with only ~2 reaching the frontier — the gap between P0 and P1 is the opportunity. ENSI components analysis (est.)
  6. You earn the AI layer; you do not skip to it. Intelligence is the capstone of Digital-Public-Infrastructure → Data → Interoperability; resource the lower layers first or you automate dysfunction at machine speed. Algorithmic State Architecture, arXiv 2503.08725
From findings to the build

These twelve findings are a dependency graph, not a menu. The remainder of the report walks them in order — the framework and landscape (Movement B), the principles, components and architectures (Movement C), the cases and recommendations (Movement D), and the risks and evidence base (Movement E) — so a nation can see not just what the agentic state is, but the sequence in which it is actually built.

The agentic race is being run on the foundations laid by the digital race — no amount of agent procurement substitutes for the identity, registries and interoperability beneath it.

ENSI, National Agentic Technology Infrastructure · synthesis 3 of 3
NATI · ENSI Report 2026The 90-Second Read
ENSI · National Agentic Technology Infrastructure10

MOVEMENT B · THE REFRAME

01 Why "buy an AI" is the wrong question


Almost every government that has put a budget line against "agentic AI" has framed it as a purchase: a chatbot here, a copilot there. That framing is the costliest mistake a nation can make — because an agent that reasons and acts on citizens is not a product you procure. It is infrastructure you provision.

For thirty years the organising metaphor of national technology was the digital service — a form turned into a web page, a registry turned into an API. The contrarian core of this report is that the next layer breaks that procurement logic. NASCIO's 2026 report frames the shift as the move from "AI that assists" to "AI that acts." Software that "perceives, reasons, and takes autonomous action within defined boundaries" is not a feature bolted onto a service. It is a new factor of production — a class of digital labour that converts policy into daily service.

And labour is provisioned, not bought off a shelf. The US GSA's USAi is a government-wide platform — chat, API and an evaluation console over multiple models — offered to every agency at no marginal cost; the EU's GenAI4EU plays the same role. Neither is "a chatbot the government bought." Each is a shared substrate on which safety, identity, evaluation and cost-control are operationalised once and inherited everywhere.

The category error, stated plainly

A chatbot answers; an agent acts. The moment software can move money, change a legal status, or take an irreversible step, it inherits the entire governance apparatus of the state — identity, authorisation, audit, liability, oversight. You cannot procure that as a SKU. The question is therefore never "which AI do we buy?" It is "what infrastructure must we provision so that agents can do useful, safe, accountable work at population scale?"

The agentic state is the digital state's canon — identity, registries, federation, interoperability, governance, oversight — extended to a worker that reasons and acts.

The spine, National Agentic Technology Infrastructure
8+US states run agentic tools in production (NASCIO 2026)
1government-wide platform — USAi — over many models, free to agencies
~10nations with the agentic P0 layer wired onto a digital foundation
200real government AI use-cases surveyed by the OECD
NATI · ENSI Report 2026Framework & Landscape
ENSI · National Agentic Technology Infrastructure11

MOVEMENT B · THE REFRAME · CONTINUED

Myth versus mechanism

Every dollar mis-spent on agentic AI can be traced to one of a handful of category errors. Each myth is intuitive, procurement-friendly and wrong. Each mechanism is the infrastructure pattern the leaders actually adopted instead.

The myth (the wrong question)The mechanism (what the leaders do instead)
Buy a chatbot for each agency. Provision one sovereign agent platform — shared model access, shared guardrails, shared observability, shared tool catalogue — that every ministry builds on (USAi, GenAI4EU). Safety and capability are inherited, not re-purchased.
AI that assists. AI that acts, under supervised autonomy — bounded action spaces, mandatory human-approval gates, a tested kill-switch. Autonomy is earned level-by-level up a maturity model, not switched on (NASCIO five-phase model).
More models = more capability. Tools + grounded data + identity = capability. An agent without tools is a text generator; an agent with the right tools, retrieving from authoritative registries, acting under a verifiable identity, is a worker (Toolformer; RAG; MCP).
Agents replace the digital state. Agents are the digital state's canon, extended. They amplify whatever they stand on — including bad data and broken process. They fail without the eID, registries, data-exchange and security beneath them.
It's an AI strategy (a PDF). It's an operating stack — compute, models, data, memory, tools, identity, orchestration, governance — provisioned, instrumented and measured by resolved outcomes, the way the EU Digital Decade reports binding KPIs.
The vanity metric

The number of agents deployed measures nothing. Count what matters: a resolved citizen outcome, a civil-servant hour returned, a cost-per-resolution that falls. Instrument the platform for task-completion, reliability under repetition, and time saved — and report it like a binding target, not a press release.

The fastest lever

Procurement, not pilots, moves adoption. The US OneGov strategy buys for "government as one customer," with agentic deals struck centrally and FedRAMP 20x fast-authorisation clearing the path. Buy agent capabilities and tokens with pre-cleared models — not bespoke per-agency contracts that re-litigate safety each time.

THE ONE-LINE TEST

If the procurement document names a product, it is the wrong question. If it names a capability provisioned once and governed for all of government, it is the right one. Agents are infrastructure, not a feature — and infrastructure is provisioned, instrumented and governed, never simply bought.

NATI · ENSI Report 2026Framework & Landscape
ENSI · National Agentic Technology Infrastructure12

MOVEMENT B · CONTINUITY

02 From digital state to agentic state


The agentic state and the digital state are not rivals. The agentic state is what you build on top of a working digital state — and it fails without one. An agent is only as good as the identity it can verify, the registry it can trust, the tool it can call, and the human who can stop it.

The Algorithmic State Architecture (arXiv 2503.08725) models the state as four stacked layers — Digital Public Infrastructure → Data → Interoperability → Intelligence — with the intelligence (agentic) layer explicitly sitting on top of the other three. You do not skip to it; you earn it. The strategic consequence is inconvenient: nations that built X-Road-class interoperability, population-scale identity and clean registries start the agentic race with half the P0 stack already in place. Those that skipped the digital foundation will find no amount of agent procurement substitutes for it.

This is why the FDA's agency-wide agentic premarket-review rollout depends on structured submission data; why Estonia's Agentic State paper lists existing X-Road interoperability and a "sovereign tech stack" as enablers. The continuity is mechanical, not rhetorical — each load-bearing principle of the digital canon has a direct agentic successor that inherits the parent's discipline rather than replacing it. Read bottom to top, the four layers are a dependency graph, not a menu.

L4 · INTELLIGENCE (agentic)

Agents that reason and act — grounded, supervised, governed. The capstone, reached only after the three layers beneath it.

L3 · INTEROPERABILITY

X-Road-style signed data exchange, base registries as single source of truth. Federation reborn as MCP + A2A.

L2 · DATA

Clean, current, machine-readable, governed records. The ground truth an agent grounds against via RAG.

L1 · DIGITAL PUBLIC INFRASTRUCTURE

eID, payments rails, connectivity, sovereign cloud, zero-trust. The trust root every transaction depends on.

THE CONTINUITY THESIS

The agentic state is not a leap past the digital state. It is the digital state's canon — identity, registries, federation, interoperability, governance, oversight — extended to a worker that reasons and acts. The order is the strategy: L1–L3 first, intelligence last.

NATI · ENSI Report 2026Framework & Landscape
ENSI · National Agentic Technology Infrastructure13

MOVEMENT B · CONTINUITY · THE INHERITANCE MAP

What each digital principle becomes

The agentic canon is the digital canon, line by line, with one world-changing clause added: the actor is now a worker that reasons and acts. Where a NATI principle has a direct digital ancestor, the synthesis flags it [↔ Pn]. The map below reads the agentic layer against the national-architecture canon.

Digital-state principle / componentIts agentic successor
eID + assurance levels (trusted identity at the root of every transaction) Agent identity + authenticated delegation — every agent has a verifiable identity; a citizen or official grants it scoped, time-bounded, revocable authority (OAuth/OIDC extended, arXiv 2501.09674).
Base registries (single source of truth) Agent-ready registries / RAG ground truth — clean, queryable systems of record the agent grounds against; what it generates is a derived view, never an alternative truth.
X-Road federation (autonomous nodes, signed point-to-point exchange) MCP + A2A — agent-to-tool (Model Context Protocol) and agent-to-agent (Agent2Agent) standards: federation reborn at the agent layer.
Interoperability framework (EIF; reusable building blocks; RIHA catalogue) Tool & agent registries — a governed catalogue of approved tools and agents with capability metadata and assurance status (AGNTCY Agent Directory).
Zero-trust security (PDP/PEP, continuously verified, least privilege) Action-scoping + kill-switch — per-agent least-privilege action sets, default-deny on irreversible operations, a tested emergency stop and rollback.
Once-only principle (don't re-ask for what the state already holds) Governed tiered memory — working, long-term and episodic case memory, governed as a regulated personal-data store (purpose limits, retention, deletion).
Central digital agency (binding reference architecture, law, targets) Central agent authority — one accountable owner of the national agent platform, under law defining permissible autonomous action and liability.
Tamper-evident logging (Estonia's KSI ledger) Agent observability + immutable logs — queryable, tamper-evident records of reasoning, tool calls and actions: the precondition for any audit.
8
digital principles with a direct agentic successor
NATI SYNTHESIS
1
added clause: a worker that reasons and acts
THE SPINE

The optimal national agentic architecture is the same sentence as the digital one — a federated, API-first, building-block platform with a trusted identity at its root and a signed exchange layer instead of a central database — now staffed by governed, grounded, supervised agents that reason and act on top of it.

Synthesis 3, The Optimal National Agentic Stack
NATI · ENSI Report 2026Framework & Landscape
ENSI · National Agentic Technology Infrastructure14

MOVEMENT B · THE FRAMEWORK

03 The 8-layer National Agentic Stack


The reference design is one stack of eight layers, sitting atop a healthy digital state. Two layers — Trust & Control (L4) and the Governance Spine (L8) — are cross-cutting. The win is not any single exotic pattern; it is having all eight wired together.

FIG B-1
LayerWhat it isTier
L8Governance Spine (cross-cutting) — binding risk framework · liability law · central agent authority · eval harness · oversightP0
L7Experience & Mission — citizen super-assistant · civil-servant copilots · policy twin · proactive life-event servicesP0
L6Orchestration — durable stateful workflows · orchestrator-worker · A2A federation · human-approval nodesP1
L5Agent Runtime — reason–act–verify loop · reflection · planning · skill libraryP0
L4Trust & Control (cross-cutting) — agent identity · delegation · action scopes · kill-switch · observability · gatewaysP0
L3Tools & Integration — MCP servers over every system · tool/agent registry · computer-use legacy bridgeP0
L2Data, Knowledge & Memory — agent-ready registries · RAG · vector+graph · context engineering · governed memoryP0
L1Compute & Models — governed frontier + sovereign models · model gateway · managed inference · cost routingP0
The 8-layer stack. Read bottom-up. ▼ Sits atop the digital state: broadband, eID, base registries, X-Road-style exchange, sovereign cloud, zero-trust — the agentic stack fails without it.

L1–L3 — the engine room, the knowledge, the reach

L1 · COMPUTE & MODELS

Role. The engine every agent runs on. P0 component: governed frontier-model access (#1). Signal: ~15–25 nations run a governed national model-access programme; USAi abstracts many models behind one gateway. P0 bedrock

L2 · DATA, KNOWLEDGE & MEMORY

Role. What agents know and remember. P0 component: authoritative, agent-ready registries (#8) + RAG layer (#10). Signal: interoperable base registries in ~40 nations; agent-ready in fewer. P0 bedrock

L3 · TOOLS & INTEGRATION

Role. How agents reach the world. P0 component: MCP tool layer (#41) — the "USB-C for AI." Signal: standard ~18 months old; early adoption in a handful of leaders. P0 bedrock

NATI · ENSI Report 2026Framework & Landscape
ENSI · National Agentic Technology Infrastructure15

MOVEMENT B · THE FRAMEWORK · CONTINUED

L4–L6 — control, runtime, orchestration

The middle of the stack is where autonomy is made survivable. L4 (Trust & Control) is cross-cutting: nothing autonomous ships before it. L5 builds the worker; L6 makes it run for hours and across agencies.

L4 · Trust & Control P0 bedrock · cross-cutting

Role. The structural member that makes autonomous action legally survivable. P0 component: agent identity (#33) + authenticated delegation (#34) + secure gateways (#40), with action-space constraints, a tested kill-switch (#36) and immutable observability (#37). Signal: production-grade national agent-identity exists in single digits of nations (est.). This is the gate the whole stack turns on.

L5 · Agent Runtime P0 bedrock

Role. How the worker itself works — the atomic unit of agency. P0 component: the standard reason–act–verify loop (#23), from ReAct and the Claude Agent SDK, with reflection (#24, Reflexion) and planning (#25) layered on. The verify step separates a reliable agent from a confident-but-wrong one. Signal: the de-facto standard in every serious build (est.).

L6 · Orchestration P1 core

Role. Turning a fragile prompt-chain into a durable process that survives crashes and hand-offs. P1 component: a stateful orchestration runtime (#27, LangGraph/AutoGen) with orchestrator-worker decomposition (#28), A2A inter-agency federation (#42) and human-approval nodes. Signal: standard tooling embedded in leading platforms; multi-agent orchestration a front-runner-only differentiator (est.).

THE LOAD-BEARING WALL

Supervised autonomy — bounded action, mandatory human gates, agent identity, authenticated delegation, full observability, a tested kill-switch — is the only pattern that appears in every credible governance source: the UK AI Playbook, NASCIO, the EU AI Act's high-risk requirements, OpenAI's practices. For a state it is not a constraint on the architecture. It is the architecture.

NATI · ENSI Report 2026Framework & Landscape
ENSI · National Agentic Technology Infrastructure16

MOVEMENT B · THE FRAMEWORK · CONTINUED

L7–L8 — mission and the governance spine

The top of the stack is where citizens meet the agentic state, and the spine that runs through all of it. L7 is built last and continuously; L8 in parallel from day one. An L7 mission deployed without L1–L6 beneath it is the most dangerous configuration in the design.

L7 · Experience & Mission P0 bedrock

Role. The visible agentic state — where policy becomes daily service. P0 component: citizen super-assistants, civil-servant copilots, a national policy twin and proactive life-event services — all bounded and human-gated. Signal: the UAE's agents-in-service vision; Singapore's NAIS 2.0; the Tony Blair Institute's "Reimagined State." Where to start: the Alan Turing Institute found ~41% of GB public-sector working time is supportable by generative AI (49% in education) — target the highest-volume, highest-burden, lowest-risk functions first.

L8 · Governance Spine P0 bedrock · cross-cutting

Role. Governance is architecture; for agents that act, it is the load-bearing wall. P0 components: a binding national risk framework (#48 — NIST AI RMF / EU AI Act / Singapore's Model Gov Framework) and a mandatory eval harness (#47 — AgentBench, GAIA, τ-bench's pass^k metric), plus a liability chain (#38) and an oversight institution (#49). Signal: a binding framework applied to agents exists in ~10–20 nations; mandatory eval regimes in very few (est.). τ-bench's finding that agents fail on repeated-trial consistency is the single most important deployment caution in the collection.

The component chips, by tier

The 50 components map onto the eight layers. The distinctively agentic P0 set is the gate almost every nation fails:

P0 frontier model access P0 agent-ready registries P0 RAG layer P0 tool-use / function calling P0 MCP tool layer P0 agent identity P0 authenticated delegation P0 secure gateways P0 human-in-the-loop gates P0 observability + logs P0 governed memory P0 eval harness P0 binding risk framework P0 national agent platform P1 sovereign inference P1 model gateway P1 orchestration runtime P1 A2A interop P2 GraphRAG P2 multi-agent P3 edge inference P3 computer-use bridge

THE BUILD SEQUENCE, IN ORDER

Pre-requisite: fix the digital state. Then L1–L2 foundation → L3–L4 trust spine (nothing autonomous ships before L4) → L5–L6 runtime → L8 governance in parallel from day one → L7 mission last and continuously. The order is the strategy.

NATI · ENSI Report 2026Framework & Landscape
ENSI · National Agentic Technology Infrastructure17

MOVEMENT B · THE GLOBAL LANDSCAPE

04 Where nations actually stand


Readiness for the agentic state is not one number; it is two. A nation's position is set by the maturity of its digital foundation (eID, registries, exchange, cloud) and its agentic capacity (platform, identity/delegation, eval, MCP). Plot the two and four quadrants appear — and almost everyone is in the wrong one.

FIG B-2
 Low agentic capacityHigh agentic capacity
Strong digital foundation Primed. Half the agentic P0 stack already built (registries, eID, exchange) but no platform, agent identity or eval yet. Estonia, Denmark/Nordics, Korea, India. Agentic frontier. Foundation + the agentic P0–P1 layer wired on. Fewer than ~10 nations; perhaps 2–3 done well. United States (federal), Singapore, UK, leading US states.
Weak digital foundation Behind the gate. Neither layer. Buying agents here yields demos, not infrastructure — you automate dysfunction at machine speed. The long tail of the ~190 economies in the GTMI. Built on sand. Mission ambition and procurement energy without clean registries or identity beneath — the most dangerous quadrant. The cautionary case for any nation that skips L1–L4.
The 2×2 readiness typology. Axes: digital-foundation maturity (vertical) × agentic capacity (horizontal). The frontier quadrant is reached only by building the foundation first — the "built on sand" quadrant is reached by skipping it.

The empirical spine beneath the map is the global connectivity and government-tech census — the floor every agentic deployment stands on:

5.5bnpeople online (68% of humanity); 2.6bn still offline — ITU 2024
96%of the world covered by a mobile-broadband network — ITU 2024
84% / 4%5G population coverage: high-income vs low-income — ITU 2024
139/194nations have a national CIRT/CERT — ITU GCI 2024
~850mpeople still lack any official ID — World Bank ID4D
198
economies scored on GovTech maturity
WORLD BANK GTMI
~40
nations with interoperable digital base registries
FOLDER 04 (est.)
<10
nations with the agentic P0 layer on a digital foundation
NATI SYNTHESIS (est.)
NATI · ENSI Report 2026Framework & Landscape
ENSI · National Agentic Technology Infrastructure18

MOVEMENT B · THE GLOBAL LANDSCAPE · READINESS

The readiness leaders — and what each is strongest at

Network readiness is the best available proxy for the digital foundation the agentic race is run on. The Network Readiness Index 2024 ranks 133 countries; its leaders are the same nations closest to the optimal agentic stack.

FIG B-3
United States#1
Singapore#2
Finland#3
Sweden / Nordics~#4
Netherlands~#6
United Kingdomtop 10
Koreatop 10
Estoniatop 20
Network Readiness Index 2024 (relative standing). Bar length = relative NRI rank among 133 countries; #1 USA, #2 Singapore, #3 Finland. The digital-readiness leaders are the pool from which the agentic frontier is drawn.

But readiness only describes the foundation. The agentic scorecard ranks each leader by the archetype it is strongest at — and names the gap that keeps even the front-runners short of the full optimal stack:

Nation / blocStrongest archetypeMain gap
US (federal)Sovereign platformCoherent governance across a fast rollout (USAi + OneGov + FDA agents).
EstoniaFederation + capstoneScaling the Agentic State framework to a production platform.
United KingdomSupervised autonomyA unifying agent platform at USAi scale (AI Playbook + Turing).
SingaporePlatform + governanceMoving from copilots to bounded autonomy (NAIS 2.0 + AI Verify).
UAEMission / experienceOpen governance + eval depth (Minister of AI; agents in visas).
EU (bloc)Federation + governanceSpeed — turning frameworks into deployed platforms (Apply AI).
Leading US statesSupervised autonomyShared platform + identity depth (8+ in production — NASCIO).
THE DECISIVE FACT

No nation has the full optimal stack. The frontier is a composite — the US's platform + Estonia's foundation + the UK's assurance discipline + Singapore's governance toolkit + the EU's binding risk law + the UAE's mission ambition. The agentic race is run on the foundations laid by the digital race.

NATI · ENSI Report 2026Framework & Landscape
ENSI · National Agentic Technology Infrastructure19

MOVEMENT B · THE SCORING LENS

05 How we rank: priority tiers


Every component and architecture in this report is placed on a five-step priority ladder. The tiers are not a wish-list; they are a dependency graph. P0 is what you build first because without it there is no safe agentic action at all.

P0 Bedrock P1 Core P2 Differentiator P3 Advanced P4 Frontier

TierDefinitionWho has it
P0 · BedrockWithout it there is no safe agentic action. Build first. 19 components.Inherited digital P0: dozens. Agentic P0: single digits to ~15 nations.
P1 · CoreDefines a serious agentic state; present in the leaders. 16 components.~5–20 — US, EU bloc, UK, Singapore, UAE, Estonia, leading US states.
P2 · DifferentiatorSeparates front-runners from followers. 9 components.~3–10 — GraphRAG, multi-agent, skill libraries, agent registries.
P3 · AdvancedTop-tier only. 4 components.~2–8 — edge inference, computer-use at scale, episodic memory.
P4 · FrontierThe fully agent-native state — intelligence as capstone.~1–3 — the state that built every layer beneath it first.

The adoption-evidence method

Agentic infrastructure is two-to-three years old as a governed national concern, so there is no ITU-grade census of it. Every adoption figure in this report is therefore one of three kinds, always labelled — and the reader should weigh them accordingly.

1 · HARD SIGNALS

Drawn directly from sources: 8+ US states in production (NASCIO 2026); the FDA's rollout; USAi & GenAI4EU exist; ITU/World-Bank census numbers. Cited to source, treated as fact.

2 · INHERITED DIGITAL COUNTS

For components the agentic state reuses from the digital state — identity, registries, cloud — counted via folder 04's indices (~40 nations with interoperable registries). Solid, but a floor: digital ≠ agent-ready.

3 · LABELLED ESTIMATES

Reasoned, order-of-magnitude bands marked (est.) — triangulated from the strategy literature. Never precise counts.

THE HONEST HEADLINE

On the foundational digital components agents depend on, dozens of nations qualify. On the distinctively agentic components — agent identity, delegation, eval harnesses, governed memory, the MCP layer, the national platform — the number with a governed, production-grade capability is in the single digits to low tens. That is precisely why the window is open.

NATI · ENSI Report 2026Framework & Landscape
ENSI · National Agentic Technology Infrastructure20
MOVEMENT C · §13

The 50 Principles of the
Agent-Native State

The convergent design canon that emerges from reading the agentic literature against the national-architecture literature — fifty rules organised into seven families (A Systemic & strategic · B Agent design & reasoning · C Models & compute · D Data & state · E Identity, trust & accountability · F Tools & integration · G Evaluation, safety & resilience) and tiered from 🟥 Foundational through 🟧 Core and 🟨 Enabling to 🟩 Frontier. They are not a menu but a dependency graph: earn the right to deploy, become a serious agentic state, separate from the pack, then reach the frontier.
50principles
7families
20🟥 foundational
4priority tiers

FAMILY A · SYSTEMIC & STRATEGIC

How the whole agentic state is shaped


The agentic state is not an AI strategy PDF or a chatbot procurement; it is a governed expansion of the state's capacity to act, built deliberately on top of a working digital foundation and resourced like headcount.

  1. Treat agents as a new factor of production, not a feature. 🟥 Stop bolting AI onto services; treat agents as digital workers that convert policy into daily service. National implication: plan agent capacity the way you plan headcount and budget, with workforce-transition planning attached.
  2. Build the agentic layer on top of a working digital state, never instead of one. 🟥 Every credible deployment sits on healthy DPI — identity, registries, data exchange, APIs; agents amplify whatever they stand on, including dysfunction. National implication: get the P0–P2 digital components healthy before scaling autonomy, or you automate dysfunction at machine speed.
  3. Agents are infrastructure, not apps — provision them centrally. 🟧 Leaders stand up shared agent platforms (GSA's USAi, the EU's GenAI4EU), not app-by-app buys. National implication: run one national platform so every ministry inherits shared models, guardrails, observability and a tool catalogue.
  4. Adopt a maturity model and sequence autonomy deliberately. 🟧 NASCIO's five phases run assistive → context-aware → task automation → stateful workflows → adaptive agents; states climbed, they did not leap. National implication: publish a maturity model and gate each rise in autonomy on demonstrated reliability at the prior level.
  5. Start where the time is, not where the hype is. 🟨 The Turing Institute found ~41% of GB public-sector working time is GenAI-supportable (49% in education). National implication: run a time-and-task analysis and target the highest-volume, highest-burden, lowest-risk functions first.
  6. Govern with binding law and a central authority. 🟥 Architecture sticks only when backed by legislation and a central agency — doubly so for agents that take legally consequential actions. National implication: name one accountable authority operating under law that defines permissible autonomous action and liability.
  7. Design for measurable outcomes, not deployed models. 🟨 Agents-deployed is a vanity metric; the unit of value is a resolved citizen outcome or a civil-servant hour returned. National implication: instrument the platform for task-completion, reliability, cost-per-resolution and time-saved, reported like binding KPIs.
  8. Sovereignty applies to agents as it does to data. 🟧 An agent reasoning over national data and acting on citizens must run under national control. National implication: hold model, orchestration and memory layers to residency and operational-autonomy controls; keep a sovereign / open-weight option for the most sensitive work.
  9. Procurement is the fastest lever you have. 🟨 OneGov "government as one customer" and FedRAMP 20x move adoption faster than any pilot. National implication: buy agent capabilities and tokens centrally, with pre-cleared models and pre-authorised tool integrations, not bespoke per-agency contracts.
  10. Plan for the AI layer to sit atop data and interoperability — earn it. 🟩 The Algorithmic State models four layers (DPI · Data · Interoperability · Intelligence) with intelligence explicitly on top. National implication: treat the agentic layer as the capstone and resource the lower layers first — an agent with no clean data is a demo, not infrastructure.
THE LOAD-BEARING SEQUENCE

Family A contains the strategy's spine: P2 and P6 are non-negotiable 🟥 foundations — a working digital state and a governing authority. Skip them and everything above is an expensive demo.

10principles · Family A
3🟥 foundational
~41%GB time GenAI-able
NATI · ENSI Report 2026The 50 Principles
ENSI · National Agentic Technology Infrastructure22

FAMILY B · AGENT DESIGN & REASONING — I

How the workers themselves work


Autonomy is a cost and a risk, spent only where dynamism is genuinely required. The atomic unit of agency is an inspectable reason–act–observe loop — and legibility is a governance asset, not a luxury.

  1. Prefer the simplest pattern that works; not everything needs an agent. 🟥 Anthropic separates workflows (predefined code paths) from agents (LLMs directing their own process) and argues for the simplest composition. National implication: make "don't use an autonomous agent where a deterministic workflow suffices" the engineering default.
  2. The reason–act–observe loop is the atomic unit of agency. 🟥 ReAct interleaves verbal reasoning, tool actions and environment observations so reasoning guides action and observation grounds reasoning. National implication: standardise reference designs on an explicit, inspectable loop rather than opaque end-to-end black boxes.
  3. Build the loop as gather-context → act → verify. 🟧 The Claude Agent SDK's cycle adds a verify step that separates a reliable agent from a confident-but-wrong one. National implication: require a verification stage (self-check, test, or second-agent review) before any consequential or irreversible action is committed.
  4. Give agents reflection and self-correction. 🟧 Reflexion's verbal self-critique stored in episodic memory improves later attempts without retraining. National implication: prefer architectures that detect and recover from their own errors — a system that catches its own mistake beats one that never doubts itself.
  5. Planning and decomposition beat single-shot generation on hard tasks. 🟨 Chain-of-Thought unlocked multi-step reasoning; Tree-of-Thoughts adds search with lookahead and backtracking. National implication: reference agents for benefits, procurement and casework should plan and decompose explicitly, with the plan exposed for audit.

A loop you cannot read is a loop you cannot govern. Interpretability is not a nicety in the public sector — it is the precondition for accountability.

Synthesis, NATI Principles (Family B)
IN SHORT

The first half of Family B fixes the shape of a public-sector agent: simplest-pattern-first, an explicit ReAct loop, a mandatory verify step, self-correction, and visible planning. Two of these five (P11, P12) are 🟥 foundational — they govern whether an agent can be inspected and trusted at all.

5 / 10Family B shown here
2🟥 foundational
P11–P15design fundamentals
NATI · ENSI Report 2026The 50 Principles
ENSI · National Agentic Technology Infrastructure23

FAMILY B · AGENT DESIGN & REASONING — II

Tools, memory and legibility


An agent without tools is a text generator; an agent with the right tools is a worker. Memory turns a chatbot into a colleague — and a colleague whose reasoning, plans and tool calls are all on the record.

  1. Tool use is a first-class capability, learned not bolted on. 🟥 Toolformer showed models learn when and how to call APIs natively; Gorilla and ToolLLM scaled it to thousands of real APIs. National implication: a platform's value is dominated by the breadth and quality of tools it can safely call — invest in the tool layer at least as much as in models.
  2. Memory turns a chatbot into a colleague. 🟧 Generative Agents' recency/importance/relevance memory stream and MemGPT's OS-style paging let an agent carry a case across sessions. National implication: design memory deliberately and govern it — an agent's long-term memory of citizens is itself a regulated data store.
  3. Accumulate reusable skills, don't re-derive them. 🟨 Voyager built an ever-growing library of self-verified, executable skills that compound — the agentic analogue of Korea's eGovFrame component libraries. National implication: catalogue, version and reuse validated agent skills across agencies rather than rebuilding the same casework agent.
  4. Match the architecture to the task's irreversibility and ambiguity. 🟧 The surveys show no single best design — the right autonomy, memory and planning depend on the task. National implication: publish a decision guide mapping task properties (reversibility, ambiguity, volume, risk) to recommended patterns, so teams choose by analysis, not fashion.
  5. Keep the agent legible: expose reasoning, plans and tool calls. 🟧 OpenAI lists legibility of agent reasoning as a baseline safety practice. National implication: mandate that production agents emit structured, retainable traces of reasoning, plans, tool calls and observations — legibility is both a debugging tool and an accountability record.
WHY THE TOOL LAYER WINS

P16 is the hinge of the whole stack: capability is dominated not by the model but by the catalogue of governed tools an agent can reach. It is why Family F (tools & integration) carries five 🟥/🟧 principles of its own — and why a national platform should out-invest its models with its connectors.

10principles · Family B total
P16–P20tools · memory · skills · legibility
1000sAPIs scaled by ToolLLM
NATI · ENSI Report 2026The 50 Principles
ENSI · National Agentic Technology Infrastructure24

FAMILY C · MODELS & COMPUTE

The engine room


Agents are only as capable as the models beneath them, and frontier capability is concentrated in a handful of providers. Model access is now the agentic equivalent of energy security — to be secured, diversified and managed, never assumed.

  1. Treat model access as a strategic national input. 🟥 Frontier capability is concentrated; the US AI Action Plan makes infrastructure a top pillar. National implication: secure governed, resilient access to frontier models (commercial and open-weight) with continuity plans for provider or jurisdictional disruption.
  2. Maintain model optionality and avoid single-model lock-in. 🟧 USAi is a unified interface over multiple leading models, not a bet on one; capability and price shift monthly. National implication: abstract the model behind a stable interface so agencies route by task, cost and assurance, and the state can switch providers without re-engineering every agent.
  3. Keep a sovereign / open-weight option for the most sensitive work. 🟨 Where data cannot leave national control, open-weight models on sovereign infrastructure are the fallback. National implication: sustain at least one in-country, controllable inference path — even if less capable — so the state is never wholly dependent on an external API for consequential action.
  4. Provision inference capacity as managed national infrastructure. 🟧 Agentic workloads are inference-heavy and bursty — one task fans out into dozens of model calls. National implication: plan inference capacity, latency and cost as a managed shared service with quotas and prioritisation, not an unmanaged per-agency cloud bill that explodes at scale.
  5. Right-size the model to the task. 🟨 Orchestrator-worker and routing patterns let a small model triage and a large model handle the hard step. National implication: bake cost-and-capability routing into the platform — at population scale, model-selection discipline is a budget line, not a detail.
  6. Engineer for latency and cost as first-order constraints. 🟧 τ-bench and GAIA show agents are slow and expensive exactly where they fan out; edge/MEC and caching matter. National implication: set SLOs for agentic interactions and design the loop (caching, parallelisation, routing) to meet them — a query that takes minutes and costs dollars will not scale to a nation.
  7. Plan the model-upgrade treadmill into operations. 🟨 Models are replaced every few months; an agent tuned to one model's quirks can regress on the next. National implication: keep an evaluation harness that re-qualifies every production agent on each model change, so upgrades are controlled rollouts, not surprises.
THE COST CLIFF

P24, P25 and P26 are one argument in three parts: agentic inference is bursty, fan-out is expensive, and naïve deployment meets a cost cliff on first contact with population scale. Managed capacity, model routing and latency SLOs are what keep the engine room affordable.

7principles · Family C
1🟥 foundational (P21)
dozensmodel calls per agent task
NATI · ENSI Report 2026The 50 Principles
ENSI · National Agentic Technology Infrastructure25

FAMILY D · DATA & STATE

What agents must know and remember


Everything an agent produces is a derived view, never an alternative truth. Authoritative registries are the ground truth; data readiness is a precondition, not a parallel workstream; and an agent's memory of citizens is regulated personal data.

  1. Authoritative registries remain the single source of truth. 🟥 Agents must act on canonical systems of record, not plausible generations. National implication: the single most valuable thing a state can do for its agents is make population, business, land and identity registries clean, current, machine-readable and queryable.
  2. Data readiness is a precondition, not a parallel workstream. 🟥 An agent automates whatever quality it is fed, at scale; governance, quality, lineage and interoperability come first. National implication: fund a data-readiness programme ahead of deployment — "agent-ready data" is the new precondition the way "machine-readable" was for open government.
  3. Ground generation in retrieval — RAG as the default. 🟧 Pairing parametric memory with a retrieval index keeps outputs factual, specific and updatable without retraining. National implication: require agents on policy, law or citizen records to retrieve from authoritative sources and cite them; ungrounded generation on official matters is a defect.
  4. Use graph knowledge stores where relationships and verifiability matter. 🟨 GraphRAG and LLM+KG enable multi-hop, verifiable reasoning that flat vector search misses — the shape of much government knowledge. National implication: build knowledge graphs over structured records as a first-class retrieval substrate, not only vector indexes.
  5. Engineer the context window deliberately. 🟧 The Context Engineering survey reframes "what goes in the window" as a formal discipline of retrieval, processing and management. National implication: specify context assembly, compression and ordering policies — context engineering is where most agent quality and cost are won or lost.
  6. Exploit long context, but don't let it replace governance. 🟨 Gemini 1.5's million-token recall lets large state be held in-context, but that is not data governance. National implication: use long context to simplify architecture where it helps, but keep authoritative retrieval and lineage — a document in context is not the source of truth.
  7. Make tool results and structured outputs first-class, persistable state. 🟧 Function-call returns and schema-constrained outputs are auditable state that flows into memory and downstream agents. National implication: standardise typed schemas so agent actions are interoperable, loggable and chainable — free text between agents is a future incident.
  8. Govern agent memory as a regulated personal-data store. 🟥 An agent's episodic and long-term memory of citizens is personal data, subject to consent, retention and deletion. National implication: apply data-protection law to agent memory explicitly — purpose limitation, retention limits, a right to deletion, with India's DEPA-style consent layer extended to agents.
NATIONAL IMPLICATION · GROUND TRUTH

Family D's spine is P28–P29–P35: registries are truth, readiness precedes deployment, and memory is regulated. Three of eight principles here are 🟥 foundational — the highest concentration of any family, because an agent that automates dirty data simply scales the error.

8principles · Family D
3🟥 foundational
1,400papers in Context survey
NATI · ENSI Report 2026The 50 Principles
ENSI · National Agentic Technology Infrastructure26

FAMILY E · IDENTITY, TRUST & ACCOUNTABILITY

No anonymous actors in production


An action you cannot attribute is an action you cannot govern. This is the densest 🟥 family in the canon: identity, scoped delegation, liability, action gating, human oversight, an interruptible kill-switch and real-time observability — six of seven are foundational.

  1. Every agent has a verifiable identity — no anonymous actors. 🟥 Agent identifiers are the first oversight measure; an unattributable action is ungovernable. National implication: extend national identity infrastructure to issue and manage agent identities, so every autonomous action traces to a specific, registered agent instance.
  2. Authenticate and authorise agents through scoped, auditable delegation. 🟥 Authenticated Delegation extends OAuth 2.0 / OIDC with agent credentials carrying least-privilege, time-bounded scope. National implication: build "delegation infrastructure" — a citizen or civil servant granting an agent specific, revocable, logged authority — as core national plumbing.
  3. Establish clear liability and principal–agent accountability. 🟥 Agency theory — information asymmetry, discretionary authority, loyalty — demands someone answerable. National implication: legislate the accountability chain so every consequential action has a named responsible human or institution; "the AI did it" must never be a valid defence.
  4. Constrain the action space and gate high-impact actions. 🟥 Default-deny for irreversible operations; require approvals for consequential ones. National implication: configure production agents with least-privilege action sets and mandatory human approval gates on anything that moves money, changes legal status, or cannot be undone.
  5. Keep a human in the loop with real authority to intervene. 🟥 The one universal guardrail — the human must review and override, not merely watch. National implication: define mandatory oversight points and escalation paths per use-case, and ensure the reviewer has the information, time and authority to actually say no.
  6. Build the kill-switch and make interruptibility testable. 🟧 Attributability and interruptibility — stop and roll back — are baseline practice. National implication: require a tested emergency stop and rollback for every production agent, and rehearse it; an autonomy you cannot halt is one you cannot deploy.
  7. Make agents observable: monitor and log everything in real time. 🟥 Real-time monitoring and activity logging are the second and third oversight measures. National implication: capture immutable, queryable logs of every agent's reasoning, tool calls and actions — the agentic analogue of Estonia's KSI tamper-evident logging.

"The AI did it" must never be a valid defence. For every consequential agent action there is a named responsible human or institution, or there is no deployment.

Synthesis, NATI Principles (Family E, P38)
7principles · Family E
6🟥 foundational
3oversight measures (Visibility)
NATI · ENSI Report 2026The 50 Principles
ENSI · National Agentic Technology Infrastructure27

FAMILY F · TOOLS & INTEGRATION

How agents reach the world


Tool calls into government systems must pass through the same secure, signed, logged discipline as X-Road traffic — never direct database access. Standardise the connector, standardise agent-to-agent coordination, and treat discovery as infrastructure.

  1. Standardise tool connection — adopt MCP as the national "USB-C for AI." 🟥 The Model Context Protocol is the emerging open standard — "build once, integrate everywhere" — the agentic counterpart to API-first and X-Road's connector. National implication: expose every registry and line-of-business system as a governed MCP server with a typed contract.
  2. Standardise agent-to-agent coordination — adopt A2A-style interoperability. 🟧 Agent2Agent's capability-advertising "Agent Cards" let agents across vendors and ministries discover and delegate. National implication: require an open interoperability standard so a citizen-services agent can securely delegate to a tax or registry agent across agency boundaries.
  3. Treat tool and agent registries / discovery as infrastructure. 🟧 A discoverable, governed catalogue of tools and agents is core plumbing — the analogue of Estonia's RIHA. National implication: operate a national registry of approved agents and tools with capability metadata, assurance status and access policy, so wiring is discovered and governed, not hard-coded.
  4. Run a control-plane orchestration runtime for long, stateful workflows. 🟧 LangGraph and AutoGen supply persistence, branching, retries and human-in-the-loop across long-running work. National implication: standardise on an orchestration layer with persistence, explicit human-approval nodes and recovery — government workflows run for hours or days, not one prompt.
  5. Use computer-use / browser agents to bridge the legacy long tail — with extra caution. 🟨 Computer use and WebArena let an agent operate human UIs directly — essential for un-API'd systems, and the highest-risk mode. National implication: permit it as a deliberate bridge under the strictest monitoring, scoping and approval — a transition, not a destination.
  6. Connect agents to the world through secure, governed gateways. 🟥 Tool calls must pass through signed, logged exchange discipline — never unmediated database access. National implication: route all agent-to-system calls through policy-enforcing gateways (authn, authz, rate-limiting, logging), preserving the digital state's exchange model when the caller is an agent.
THE FEDERATION PRINCIPLE, REBORN

Family F is X-Road for agents. P43 (MCP) is the connector, P44 (A2A) is cross-agency federation, P45 is the RIHA-style registry, and P48 is the signed gateway. The digital state's interoperability canon is not replaced — it is extended to a caller that reasons and acts.

6principles · Family F
2🟥 foundational (P43, P48)
MCP"USB-C for AI"
NATI · ENSI Report 2026The 50 Principles
ENSI · National Agentic Technology Infrastructure28

FAMILY G · EVALUATION, SAFETY & RESILIENCE

Governance is the load-bearing wall


The smallest family carries the two most absolute rules: no deployment without an evaluation harness, and a binding national risk-management spine. For agents that act, governance is not paperwork — it is architecture.

  1. No deployment without an evaluation harness — and continuous re-evaluation. 🟥 AgentBench (competence), GAIA (real-world assistance) and τ-bench (policy adherence, the pass^k reliability metric) measure capability and consistency; τ-bench's repeated-trial failures are the single most important deployment caution in the collection. National implication: require every production agent to pass a documented eval suite — capability, reliability under repetition, safety, policy-adherence — re-qualified continuously and on every model change.
  2. Adopt a national risk-management spine and a trustworthy-AI bar. 🟥 NIST AI RMF (Govern · Map · Measure · Manage) with its GenAI Profile, the EU AI Act's risk tiers, and Singapore's Model AI Governance Framework with AI Verify define the master control structure — validity, safety, security, accountability, transparency, explainability, privacy and fairness. National implication: adopt one framework as the binding national spine, map every system to its risk tier, and make conformity a precondition of deployment.
THE DEPLOYMENT CAUTION

τ-bench's finding that agents frequently fail on repeated-trial consistency — passing once, then failing the same task on retry — is why P49 demands reliability-under-repetition, not just a one-shot benchmark. An agent that works in the demo and fails the fourth caller is not deployable.

How the 50 compose: the agentic build sequence

The principles are a dependency graph, read as a sequence of four tiers.

🟥 Foundational — earn the right to deploy20
🟧 Core — a serious agentic state18
🟨 Enabling — separate from the pack11
🟩 Frontier — the maturing edge1

The agentic state is not a leap past the digital state — it is the digital state's canon (identity, registries, federation, interoperability, governance, oversight) extended to a worker that reasons and acts.

Synthesis, NATI Principles (the spine)
2principles · Family G
2🟥 foundational
50 / 50principles covered
NATI · ENSI Report 2026The 50 Principles
ENSI · National Agentic Technology Infrastructure29
MOVEMENT C · §15

The 50 Components — A Bill of Materials for the Agentic State

Seven layers stack from the engine room to the governance spine: compute & models, data & knowledge, memory & state, agent runtime, identity & trust, tools & integration, and governance & people. Each component carries a priority tier — P0 Bedrock through P4 Frontier — and an honest count of how many nations actually have it. Read down the tiers and a single finding emerges: the leaderboard is short, and it gets shorter fast.

MOVEMENT C · §15 · THE 50 COMPONENTS

L1 Compute & Model Substrate — The Engine Room


An agent is only as capable as the model it runs on and the silicon that serves it. Layer 1 is the engine room: governed access to frontier reasoning, a sovereign fallback for data that cannot leave the country, and the burst capacity that population-scale agent fan-out devours. Two of its seven components are P0 — without model access and serving capacity there are simply no agents.

#ComponentTierWhat it isAdoption
1Frontier model access (governed)P0Assured, governed access to frontier LLMs — the reasoning engine every agent runs on~15–25 nations (est.)
2Sovereign / open-weight inference pathP1In-country, controllable inference for data that cannot leave national control~10–15 nations (est.)
3Managed inference capacityP0GPU/accelerator serving capacity for inference-heavy, bursty agent workloads~20–30 nations (est.)
4Model gateway / abstraction layerP1A stable internal API hiding specific providers — optionality, no lock-in~10–20 nations / inst. (est.)
5Cost-and-capability routingP2Sends each step to the cheapest model that can do itFew national platforms (est.)
6Inference observability & FinOpsP2Telemetry on token spend, latency, throughput per agent, with quotasMinority of platforms (est.)
7Edge / low-latency inferenceP3Inference pushed to the edge for latency-sensitive, data-local tasksHandful of nations (est.)
DEEP DIVE · #1 FRONTIER MODEL ACCESS · P0

Effectively every nation can buy API access; almost none has governed access — central contracts plus a routing layer so agencies draw only on vetted models. The flagships are the US USAi platform and the EU's GenAI4EU; governed national programmes exist in an estimated 15–25 nations, clustered among the US, UK, Singapore, the UAE and the Gulf, and China. The gap between "can buy" and "governs" is the first place the curve bends.

DEEP DIVE · #3 MANAGED INFERENCE CAPACITY · P0

Agent tasks fan out into many model calls, so serving capacity — not model quality — is the binding constraint at scale. National AI-compute initiatives (the US AI Action Plan infrastructure pillar, EU, UK, Gulf, India, Japan, Korea) put roughly 20–30 nations in the frame, but explicit capacity reservation tuned for bursty agentic load remains rarer than the headline count suggests.

FIG 15.1
#3 Managed inference capacity~25
#1 Frontier model access (gov.)~20
#4 Model gateway layer~15
#2 Sovereign inference path~12
#5 Cost/capability routing~6
#7 Edge inference~4
Layer 1 adoption — approximate nation-counts (est.). Midpoints of the synthesis bands; all figures order-of-magnitude, not a census. The engine room is the most widely held layer — and even here the governed, routed variants thin out fast.
NATI · ENSI Report 2026The 50 Components
ENSI · National Agentic Technology Infrastructure31

MOVEMENT C · §15 · THE 50 COMPONENTS

L2 Data & Knowledge Substrate — What Agents Know


Agents automate whatever data quality they are fed; bad registries plus agents produce errors at machine speed. Layer 2 is the substrate of ground truth — authoritative registries, a data-readiness programme, and the retrieval pipeline that grounds every official answer in a citable source. Three of its nine components are P0: registries, governance, and RAG are not parallel work, they are prerequisites.

#ComponentTierWhat it isAdoption
8Authoritative, agent-ready registriesP0Clean, current, machine-readable systems of record — the ground truth agents act on~40 base; fewer agent-ready (est.)
9Data-readiness & governance programmeP0Quality, lineage, access rights, stewardship over data agents consume~30–50 nations (est.)
10Retrieval-Augmented Generation (RAG) layerP0Pipeline grounding agent output in authoritative sourcesMost serious pilots (est.)
11Vector knowledge storeP1Dense embedding index for semantic search over unstructured corporaStandard where RAG runs (est.)
12Graph knowledge store / GraphRAGP2Entity-relationship graphs for multi-hop, verifiable reasoningMinority of platforms (est.)
13Context-engineering layerP1Discipline of assembling, compressing, ordering what enters the context windowFormalised in few (est.)
14Long-context capabilityP2Million-token models holding large state in-contextAny frontier-model user
15Structured-output / schema layerP1Typed, schema-constrained outputs that are auditable and chainableStandard in mature builds (est.)
16Consent / data-empowerment for agentsP2Consent infrastructure governing what data agents may access & rememberDEPA-style ~15; agent-aware few (est.)
DEEP DIVE · #8 AGENT-READY REGISTRIES · P0

Interoperable digital base registries exist in roughly 40 nations, but agent-ready — clean, exposed via governed APIs, with lineage an agent can cite — is a higher bar that fewer clear. This is the inheritance thesis in miniature: the nations that built the digital foundation start the agentic race with this component half-built; the nations that did not will find that buying agents on dirty registries yields confident, fast, wrong answers.

DEEP DIVE · #10 RAG LAYER · P0

Ungrounded generation on official matters is a defect, not a feature. The RAG pipeline — chunk, embed, index, rewrite, re-rank — wires an embedding model and a vector index into the agent loop so every answer traces to an authoritative source. It is ubiquitous in private deployments and a standard component of most serious national pilots; the differentiator is no longer having RAG but governing its corpus and provenance.

FIG 15.2
#9 Data-readiness programme~40
#8 Agent-ready registries~32
#16 DEPA-style consent~15
#11 Vector store / RAG (serious)~12
#12 GraphRAG~7
#16 Agent-aware consent~4
Layer 2 adoption — approximate nation-counts (est.). Inherited digital-state counts (registries, governance, consent) are widely held; the distinctively agentic refinements — agent-ready data, agent-aware consent — collapse toward single digits.
NATI · ENSI Report 2026The 50 Components
ENSI · National Agentic Technology Infrastructure32

MOVEMENT C · §15 · THE 50 COMPONENTS

L3 Memory & State — What Agents Remember


Memory turns an agent from a one-shot tool into an institution that carries a case across sessions and learns from experience. Layer 3 climbs from the context-window scratchpad every agent has, through OS-style paging and episodic streams, to the governed memory store the law actually requires — because an agent's memory of citizens is personal data. Two P0s anchor it: working memory, and the regulated store that makes the rest lawful.

#ComponentTierWhat it isAdoption
17Short-term / working memoryP0The context window as the agent's active scratchpad for the current taskUniversal to any agent
18Long-term memory with pagingP1OS-style tiered memory (MemGPT) for an unbounded effective horizonLeading platforms (est.)
19Episodic memoryP2Time-stamped memory stream retrieved by recency × importance × relevanceA minority (est.)
20Semantic memory & reflectionP2Distilled higher-level facts synthesised from raw episodes, with forgettingResearch→early prod. (est.)
21Governed memory store (regulated data)P0Agent memory treated as regulated personal data — purpose limits, retention, deletionOperationalised in very few (est.)
22Skill / workflow libraryP2Catalogue of validated, reusable agent skills shared across agenciesA few leaders (est.)
DEEP DIVE · #21 GOVERNED MEMORY STORE · P0

The moment an agent remembers a citizen, its memory becomes a regulated personal-data store — subject to purpose limitation, retention limits and deletion rights, with audit over every read and write. The governance literature recognises this as a requirement; almost no nation has operationalised it. It is the quietest P0 in the catalogue and one of the most consequential: an ungoverned memory is a standing data-protection breach that compounds with every interaction.

DEEP DIVE · #22 SKILL / WORKFLOW LIBRARY · P2

A versioned national repository of approved agent workflows is the agentic eGovFrame — build once, reuse many. Following Voyager's skill-library pattern, validated skills become shared assets across ministries rather than re-implemented per agency. It is nascent: a few leaders are assembling catalogues, but most builds still reinvent the same casework loop in every department, paying the integration cost repeatedly.

FIG 15.3
#17 Working memoryuniv.
#18 Long-term memory (paging)~12
#19 Episodic memory~7
#22 Skill / workflow library~5
#20 Semantic memory / reflection~4
#21 Governed memory store~3
Layer 3 adoption — approximate nation-counts (est.). Working memory is free with every agent; the governed, regulated memory store the law demands is the rarest piece on the page — a P0 honoured in principle and skipped in practice.
NATI · ENSI Report 2026The 50 Components
ENSI · National Agentic Technology Infrastructure33

MOVEMENT C · §15 · THE 50 COMPONENTS

L4 Agent Runtime & Patterns — How Agents Work


Layer 4 is the mechanics of agency: the reason–act–verify loop, the tool-calling that turns a text generator into a worker, the approval gates that hold a human in command of consequential action, and the orchestration runtime that carries a workflow over hours or days. Four P0s define it — the loop, tool use, human-in-the-loop gates — and they are where "demo" becomes "infrastructure."

#ComponentTierWhat it isAdoption
23Standard agent loop (reason–act–verify)P0Explicit, inspectable gather→act→verify loop — the atomic unit of agencyDe-facto standard (est.)
24Reflection / self-correctionP1Generate–critique–revise loops that catch the system's own errorsLeading deployments (est.)
25Planning & task decompositionP1Decompose a goal into an ordered, adaptable plan exposed for auditStandard in complex agents (est.)
26Tool-use / function-callingP0Deciding when/how to call external tools and folding results backUniversal; safe use is the gap
27Orchestration runtime (stateful)P1Control plane for long-running, multi-step workflows with persistence & recoveryLeading platforms (est.)
28Multi-agent orchestrationP2Manager agent decomposing work and delegating to specialised workersFront-runners only (est.)
29Human-in-the-loop approval gatesP0Mandatory review/override before consequential or irreversible actions8+ US states; leaders (NASCIO 2026)
30Verification / self-test stepP1A check or second agent before committing an actionLeading practice (est.)
31Computer-use / browser-agentP3Agents operating human UIs directly for un-API'd legacy systemsPilots in a few nations (est.)
32Workflow-vs-agent decision standardP2Published guide mapping task properties to the simplest sufficient patternRare as national policy (est.)
DEEP DIVE · #29 HUMAN-IN-THE-LOOP GATES · P0

The universal guardrail across the UK Playbook, NASCIO and the EU AI Act: mandatory review and override before any irreversible action, implemented as approval nodes in the runtime. One of the few agentic P0s with a hard signal — 8+ US states run agentic tools with these gates operationalised (NASCIO 2026).

DEEP DIVE · #23 STANDARD AGENT LOOP · P0

An explicit gather-context → act → verify loop (ReAct, the Claude Agent SDK) is the atomic unit of agency and the basis of interpretability: an inspectable loop is an auditable agent. It is de-facto standard in every serious build — so the differentiators live upstairs in orchestration and downstairs in trust.

FIG 15.4
#26 Tool-use / function-callinguniv.
#29 Human-in-the-loop gates8+
#27 Orchestration runtime~10
#28 Multi-agent orchestration~6
Layer 4 adoption — approximate counts. #29 carries a NASCIO 2026 hard signal (8+ US states); the rest are est. midpoints. The loop and tool-use are universal; governed, multi-agent variants are front-runner territory.
NATI · ENSI Report 2026The 50 Components
ENSI · National Agentic Technology Infrastructure34

MOVEMENT C · §15 · THE 50 COMPONENTS

L5 Identity, Trust & Control for Agents


This is the load-bearing layer of the agentic state — the agentic extension of eID. An unattributable action is an ungovernable one, so every agent needs a verifiable identity, a scoped and revocable mandate, a least-privilege action space, an immutable log, and a named human on the legal hook. Six of its eight components are P0, more than any other layer, and they are exactly where adoption is thinnest.

#ComponentTierWhat it isAdoption
33Agent identity serviceP0Verifiable identities for every agent instance — attribution of actionSingle digits of nations (est.)
34Authenticated delegation infrastructureP0Scoped, time-bounded, revocable, logged authority — OAuth/OIDC for agentsVery few nations (est.)
35Action-space constraint / least-privilegeP0Per-agent action sets, default-deny on irreversible operationsInconsistently done (est.)
36Kill-switch & interruptibilityP1A tested emergency stop and rollback for every production agentTested in few (est.)
37Agent observability & immutable loggingP0Real-time monitoring + immutable logs of reasoning, tool calls, actionsLeaders (est.)
38Liability & accountability frameworkP0Legal chain naming a responsible human for every consequential actionBinding regimes in few (est.)
39Guardrails & policy-enforcement layerP1Content/safety filters, domain rules, output provenance on agent actionsAdopted by leaders (est.)
40Secure agent-to-system gatewaysP0Policy-enforcing gateways between agents and systems — the X-Road disciplineInherited; agent-specific in leaders (est.)
DEEP DIVE · #33 AGENT IDENTITY SERVICE · P0

Issuing and managing verifiable identities for every agent instance is the foundation of the whole trust layer — an action no one can attribute is an action no one can govern. The fix is to extend existing national identity infrastructure to issue agent identities. It is a recognised requirement, yet production-grade national agent-identity exists in only single digits of nations. This is the single steepest fall in the catalogue.

DEEP DIVE · #34 AUTHENTICATED DELEGATION · P0

The agentic extension of eID: the ability for a citizen or civil servant to grant an agent a scoped, time-bounded, revocable, logged authority — OAuth/OIDC extended for agents. It is the technical basis of "acting on behalf of," and it is emerging in only very few nations as actual plumbing. Without it, every agent action is either over-privileged or unattributable; with it, delegation becomes governable infrastructure.

FIG 15.5
#40 Agent-to-system gateways~11
#37 Observability & logging~9
#39 Guardrails layer~8
#33 Agent identity service~6
#34 Authenticated delegation~4
#38 Liability framework (binding)~3
Layer 5 adoption — approximate nation-counts (est.). The EU AI Act is the leading liability instrument. This is the thinnest layer in the catalogue: six P0 controls, none widely held, and the distinctively agentic ones — identity, delegation, binding liability — in low single digits.
NATI · ENSI Report 2026The 50 Components
ENSI · National Agentic Technology Infrastructure35

MOVEMENT C · §15 · THE 50 COMPONENTS

L6·L7 Tools & Integration · Governance, Eval & People


The final two layers reach outward and upward. Layer 6 is how agents touch the world — the MCP tool layer (the national "USB-C for AI"), agent-to-agent federation, registries, and the unifying national platform every ministry should inherit. Layer 7 is the governance spine: the mandatory eval harness, the binding risk framework, the oversight institution, and the workforce that works alongside agents. Five P0s span the two.

#ComponentTierWhat it isAdoption
41Model Context Protocol (MCP) tool layerP0Every system exposed as a governed MCP server — the national "USB-C for AI"Handful of leaders (est.)
42Agent-to-agent interoperability (A2A)P1Open standard letting agents discover & delegate across vendors & ministriesNascent; pilots (est.)
43Agent & tool registry / discoveryP1National registry of approved agents/tools with capability & assurance metadataVery few registries (est.)
44Tool-calling correctness & scale toolingP2Map intent to correct API calls across thousands of services, no hallucinationFront-runners (est.)
45Integration to legacy via computer-use bridgeP3Governed computer-use as a monitored bridge to never-API'd systemsEarly pilots (est.)
46National agent platform (unifying substrate)P0Shared platform binding models, tools, memory, guardrails, observability~5–15 nations (est.)
47Agentic evaluation harnessP0Required eval suite — capability, reliability (pass^k), safety, policy adherenceMandatory regimes very few (est.)
48National AI risk-management framework (binding)P0Master control spine — NIST AI RMF / EU AI Act tiers / Singapore MGFBinding & applied ~10–20 (est.)
49Oversight & assurance institutionP1A body & process for human oversight, escalation, independent assuranceA minority of nations (est.)
50Workforce & capability programmeP1Training civil servants to work with agents; role-transition planning~20–40 nations (est.)
#41 MCP TOOL LAYER · P0

Every registry and line-of-business system exposed as a governed MCP server with a typed contract — the national USB-C for AI. Platform value is dominated by the breadth of tools agents can safely call. The standard is ~18 months old; national adoption sits in a handful of leaders.

#46 NATIONAL AGENT PLATFORM · P0

The shared substrate binding models, tools, memory, guardrails and observability so every ministry inherits safety rather than re-inventing it — government-as-a-platform, agentic. Real platforms exist in ~5–15 nations: USAi, GenAI4EU, the UK, Singapore, the UAE, Estonia.

#47 EVAL HARNESS & #48 RISK FRAMEWORK · P0

τ-bench shows agents fail on consistency, so no eval means no deploy — yet mandatory regimes exist in very few nations. The binding risk framework (NIST AI RMF, EU AI Act, MGF) is the load-bearing wall; applied to agents, live in ~10–20.

FIG 15.6
#50 Workforce programme~30
#48 Binding risk framework~15
#46 National agent platform~10
#41 MCP tool layer~5
Layers 6 & 7 adoption — approximate nation-counts (est.). Workforce and risk frameworks inherit broad reach; the distinctively agentic P0s — MCP, the eval harness — sit at the bottom, the gate the rest of the stack waits behind.
NATI · ENSI Report 2026The 50 Components
ENSI · National Agentic Technology Infrastructure36

MOVEMENT C · §15 · THE CENTRAL FINDING

The Adoption Curve — Where Nations Fall Off


Read the 50 components down their priority tiers and the catalogue resolves into one shape: a steep descending curve. On the foundational digital pieces agents reuse, dozens qualify. On the distinctively agentic components — agent identity, delegation, eval harnesses, registries — the count with a governed, production-grade capability is single digits to low tens, falling toward one or two at the frontier.

19
P0 Bedrock components
NATI COMPONENTS DOC
<10
nations with the agentic P0 layer wired together
SYNTHESIS (EST.)
8+
US states running agentic tools in production
NASCIO 2026
FIG 15.7
P0 Bedrock — digital prerequisitesdozens
P0 Bedrock — distinctively agentic~<15
P1 Core — "serious agentic state"~5–20
P2 Differentiator~3–10
P3 Advanced~2–8
P4 Frontier — agent-native state~1–3
The agentic adoption curve. Approximate nation-counts by tier (all est. except the NASCIO hard signal). The split inside P0 is the whole story: the inherited digital bedrock is broadly held; the distinctively agentic bedrock collapses to under ~15 nations and keeps falling.
STEEPER THAN THE DIGITAL STATE

For digital infrastructure, nearly every nation held the P0 layer and the drop-off came later, at P2–P4. For agentic infrastructure it starts almost immediately: the seven distinctively agentic P0 components exist together in fewer than ~10 nations. The curve does not taper — it cliffs.

The optimal agentic architecture is not exotic — it is the agentic P0–P1 components built and wired onto a healthy digital state. Almost no one has done it yet.

NATI Synthesis, B · The 50 Components
NATI · ENSI Report 2026The 50 Components
ENSI · National Agentic Technology Infrastructure37

MOVEMENT C · §15 · THE CENTRAL FINDING

Why the Window Is Open


The short leaderboard is not a counsel of despair — it is the strategic opening. Because the agentic P0–P1 layer is held by so few, a nation that already built the digital foundation can leap to the front of a race barely two years old. The table restates the curve as a stepped ledger: how many nations clear each tier, and what the tier gates.

TierComp.Nations (est.)Who has them & what the tier gates
P0 Bedrock19digital: dozens
agentic: <~15
The agentic P0 — identity, delegation, eval harness, governed memory, MCP, platform, binding risk framework — is the real gate
P1 Core16~5–20The "serious agentic state" band: US, the EU bloc, UK, Singapore, UAE, Estonia
P2 Differentiator9~3–10GraphRAG, multi-agent orchestration, skill libraries, registries
P3 Advanced4~2–8Edge inference, computer-use at scale, episodic memory in production
P4 Frontier~1–3The fully agent-native state — folded into P3

The inheritance dividend

The nations that built the digital foundation — Estonia, Singapore, the UAE, the UK, leading US states — start the agentic race with components 8 (registries), 9 (data governance), 16 (consent), 40 (secure gateways) and 46 (national platform) already half-built. The agentic state is the digital state's canon — identity, registries, federation, interoperability, governance, oversight — extended to a worker that reasons and acts. The nations that did not face the harder lesson: buying agents without the foundation yields demos, not infrastructure — a fast, confident way to be wrong at population scale.

5
inherited components leaders already hold (#8, 9, 16, 40, 46)
NATI SYNTHESIS
METHODOLOGY NOTE

Hard figures — 8+ US states in production (NASCIO 2026), the FDA agency-wide rollout, USAi and GenAI4EU — are cited to source. Inherited digital-state counts trace to the digital components catalogue. Every (est.) figure is a reasoned, order-of-magnitude band: no global census of national agentic infrastructure yet exists, and any single precise number should be treated with caution.

NATI · ENSI Report 2026The 50 Components
ENSI · National Agentic Technology Infrastructure38
MOVEMENT C · §19

The 10 Best Agentic
Architecture Types

Ranked by how load-bearing each pattern is for a national-scale, accountable deployment — a blend of how decisive it is for governed autonomy, how well-evidenced it is in the field, and how sharply it separates a thinking state from a chatbot procurement. The ranking closes on the optimal reference design: the 8-layer National Agentic Technology Stack that braids all ten onto a working digital foundation.
#1 Sovereign Agent Platform #2 Grounded RAG-Agent #3 Orchestrator–Worker #4 Supervised Autonomy #5 A2A Federation #6 MCP Tool-Augmented #7 Stateful Workflow #8 Reflective Agent #9 Memory-Augmented #10 Computer-Use Bridge

PART 1 · THE RANKED ARCHETYPES — 1 OF 6


The ten types are not rivals but emphases; a mature national stack braids several together. They are presented below to a single anatomy — shape, why it ranks, trade-off, exemplars and a tier/role chip — so the run reads as one comparable family.

FIG 19.1 · RANK LADDER
01 Sovereign Agent Platform97
02 Grounded RAG-Agent93
03 Orchestrator–Worker88
04 Supervised Autonomy86
05 A2A Federation80
06 MCP Tool-Augmented78
07 Stateful Workflow73
08 Reflective Agent69
09 Memory-Augmented64
10 Computer-Use Bridge52
Load-bearing score (illustrative / est.). A blended index of how decisive each pattern is for governed national autonomy, how well-evidenced it is, and how sharply it separates a real agentic state from a chatbot. Magenta = load-bearing P0/P1 cornerstones.

01 The Sovereign Agent Platform P0 bedrock · where you build

SHAPE

A single, centrally-provisioned national platform giving every agency shared model access (behind a gateway), shared tools (via MCP), shared memory, guardrails, observability and an eval harness — at no marginal cost. Ministries build agents on the platform rather than buying them piecemeal.

WHY IT RANKS #1

The agentic successor to Singapore's SGTS and the UK's GDS — the single highest-leverage move a state can make. It is where every other principle is operationalised once: safety, identity, evaluation, cost control. Without it a nation gets a thousand ungoverned pilots; with it, governed infrastructure.

TRADE-OFF

Needs a powerful, well-funded central authority and the political capital to make the platform mandatory. Weaker fit for highly federal states unless paired with the federation pattern (#5).

NATI · ENSI Report 2026The 10 Architectures
ENSI · National Agentic Technology Infrastructure40

THE RANKED ARCHETYPES — 2 OF 6


02 The Grounded RAG-Agent P0 bedrock · how you stay correct

SHAPE

Every consequential agent is a reason–act loop whose every factual claim and action is grounded in retrieval from authoritative sources — base registries, law, policy, case files — through a RAG layer (vector + graph), with citations and structured outputs. Ungrounded generation on official matters is treated as a defect.

WHY IT RANKS #2

For a state, correctness and provability outrank cleverness. The foundational RAG work and the RAG survey establish grounding as the antidote to hallucination; OECD/GPAI makes authoritative government data the ground truth agents must act on. This is what makes an agent's output defensible — it can show its sources.

TRADE-OFF

Quality is bounded by data readiness; on a messy registry the RAG-agent faithfully retrieves garbage. It forces the correct but expensive discipline of data-readiness first.

For a state, correctness and provability outrank cleverness. A government that deploys ungrounded agents on official matters is automating unaccountability.

Synthesis C, Ten Agentic Architectures & the Optimal Stack

03 The Orchestrator–Worker Multi-Agent System P1 core · how agents are built

SHAPE

A manager/orchestrator agent decomposes a complex task, delegates sub-tasks to specialised worker agents, and synthesises their results — the agentic analogue of a case team. Both Anthropic's Building Effective Agents and OpenAI's Practical Guide name this the primary multi-agent pattern.

WHY IT RANKS #3

Government work is cross-domain — one benefit claim may touch identity, tax, health and housing. Specialised agents, each with its own tools and policy scope, cooperate under one coordinator, scaling to casework a single agent cannot hold. It maps naturally onto the federated structure of government itself.

TRADE-OFF

Coordination overhead, error propagation between agents, harder debugging. Anthropic's own guidance: don't reach for multi-agent until a single agent provably cannot do the job.

NATI · ENSI Report 2026The 10 Architectures
ENSI · National Agentic Technology Infrastructure41

THE RANKED ARCHETYPES — 3 OF 6


04 The Human-in-the-Loop Supervised-Autonomy Architecture P0 bedrock · how you stay safe

SHAPE

Agents operate autonomously within bounded action spaces, but every consequential or irreversible action passes through a mandatory human approval gate, with escalation paths, full observability and a tested kill-switch. Autonomy is earned level-by-level up a maturity model.

WHY IT RANKS #4

The only pattern that appears in literally every credible governance source — the UK AI Playbook, NASCIO's maturity model, the EU AI Act's high-risk requirements, OpenAI's governance practices, the oversight paper (arXiv 2506.04836). For a state, supervised autonomy is not a constraint on the architecture; it is the architecture — what makes autonomous action politically and legally survivable.

TRADE-OFF

Human gates cap throughput and can decay into rubber-stamps when the reviewer lacks time, information or authority. The design challenge is meaningful oversight, not theatrical oversight.

05 The Federated Inter-Agent Mesh (A2A / Internet of Agents) P1 core · how agents reach each other

SHAPE

Each agency runs its own agents over its own systems; agents discover and delegate to each other across boundaries via an open agent-to-agent standard (Agent2Agent's capability-advertising "Agent Cards"; AGNTCY's federated Agent Directory). No central monolith — a mesh of autonomous, interoperating agents, mirroring X-Road's federation at the agent layer.

WHY IT RANKS #5

The agentic reincarnation of the #1 digital archetype (X-Road federation). It preserves agency autonomy and data ownership while enabling cross-agency action — a citizen-services agent securely delegating a sub-task to a tax or registry agent it discovered. For large, federal, sovereignty-conscious states it is the natural shape.

TRADE-OFF

Nascent standards (A2A donated to the Linux Foundation only in 2025), immature security/trust models for cross-agent delegation, and harder end-to-end accountability across a mesh.

NATI · ENSI Report 2026The 10 Architectures
ENSI · National Agentic Technology Infrastructure42

THE RANKED ARCHETYPES — 4 OF 6


06 The MCP-Connected Tool-Augmented State P1 core · how agents reach the world

SHAPE

Every government registry, database and line-of-business system is exposed as a governed MCP server with a typed contract; agents are MCP clients that integrate through one standard — "build once, integrate everywhere." The state's API estate becomes a uniform, discoverable tool-surface, behind policy-enforcing gateways.

WHY IT RANKS #6

The agent surveys and the tool-use literature (Toolformer, Gorilla, ToolLLM) are unanimous: an agent's value is dominated by the tools it can safely call. MCP is the emerging open standard that turns the whole government estate into agent-callable tools without bespoke wiring — the agentic extension of the digital state's API-first principle, and the substrate beneath every other pattern.

TRADE-OFF

Exposing systems as tools multiplies the attack surface; demands rigorous gateway-level auth, scoping, rate-limiting and logging. The standard is young and evolving.

07 The Stateful Orchestrated Workflow P1 core · durable agentic process

SHAPE

Long-running processes — a visa case, a procurement, a benefits review — are modelled as durable, stateful graphs (LangGraph-style): nodes are agents/tools/decision points, state is checkpointed and persisted, and the workflow survives failure, supports retries and interposes human-approval nodes. Process, not chat, is the unit.

WHY IT RANKS #7

Real public-service work runs for hours, days or weeks and must survive crashes, restarts and hand-offs. Orchestration runtimes provide the persistence, branching and recovery that turn a fragile prompt-chain into a durable business process — the pattern that makes agents reliable enough for casework rather than just Q&A.

TRADE-OFF

Engineering complexity; modelling government processes as graphs is real work and can ossify if not maintained.

NATI · ENSI Report 2026The 10 Architectures
ENSI · National Agentic Technology Infrastructure43

THE RANKED ARCHETYPES — 5 OF 6


08 The Reflective / Self-Correcting Agent P2 differentiator · evaluator-optimizer

SHAPE

Agents critique and revise their own work before committing it — a generate→critique→revise loop, often with a separate evaluator agent — and store lessons from failures in episodic memory (Reflexion's verbal reinforcement learning; the evaluator-optimizer workflow). The verify stage of the Claude Agent SDK loop is the production form.

WHY IT RANKS #8

In public service, a system that catches its own mistake is worth more than one marginally more accurate that never doubts itself. Reflection (arXiv 2303.11366) measurably improves reliability without retraining and produces an auditable trail of self-correction — valuable for accountability.

TRADE-OFF

Extra latency and cost per task; reflection can entrench a confident error if the critic shares the generator's blind spot — hence prefer an independent evaluator.

09 The Memory-Augmented Long-Horizon Agent P2 differentiator · governed memory

SHAPE

Agents carry persistent, tiered memory — working memory (context), long-term memory with OS-style paging (MemGPT), episodic memory (the recency–importance–relevance stream of Generative Agents) and consolidated semantic memory — all governed as regulated personal data. The agent carries a case across sessions and accumulates institutional knowledge.

WHY IT RANKS #9

Statelessness is the difference between a chatbot and a colleague. For recurring relationships — a citizen's multi-year benefits history, an ongoing investigation — memory is what makes the agent useful and humane. The memory literature gives the architecture; the governance literature insists it be regulated.

TRADE-OFF

Agent memory of citizens is a serious privacy and security liability — it must be purpose-limited, retention-bounded, deletable and audited, or it becomes a surveillance apparatus. The architecture is inseparable from its governance.

NATI · ENSI Report 2026The 10 Architectures
ENSI · National Agentic Technology Infrastructure44

THE RANKED ARCHETYPES — 6 OF 6


10 The Computer-Use Legacy Bridge P3 transition · highest-risk integration

SHAPE

For the long tail of government systems that will never expose an API, agents operate the human user interface directly — reading the screen, clicking, typing (Anthropic computer use; WebArena's web-task environment) — inside a tightly sandboxed, monitored, strictly-scoped environment, as a deliberate transition mechanism.

WHY IT RANKS #10

Every state has decades of un-API'd legacy software; without a bridge those systems are invisible to agents and the agentic state stalls at its own legacy boundary. Computer-use is the pragmatic — and genuinely powerful — bridge. It ranks last because it is also the highest-risk integration mode: an agent with general UI control and weak scoping is the most dangerous configuration in this entire document.

TRADE-OFF

Brittle, slow, and dangerous if under-governed; a destination only by mistake. Treat strictly as a transition while the underlying systems are API-enabled or replaced.

META-OBSERVATION · THE TEN ARE A BRAID, NOT A SHORTLIST

As with the digital archetypes, the top patterns are complementary emphases, not rivals. #1 (sovereign platform) is where you build; #4 (supervised autonomy) is how you stay safe; #2 (grounded RAG) is how you stay correct; #6 (MCP tools) and #5 (A2A federation) are how agents reach the world and each other; #3 / #7 / #8 / #9 are how the agents themselves are built; #10 is how you bridge the legacy past. The optimal stack on the pages that follow is the braid of all ten.

4P0 bedrock patterns (#1,#2,#4,+identity)
10archetypes, all load-bearing in a mature braid
<10nations with the P0–P1 layer actually wired
NATI · ENSI Report 2026The 10 Architectures
ENSI · National Agentic Technology Infrastructure45

PART 2 · THE OPTIMAL NATIONAL AGENTIC STACK · 2026


The Reference Design — 8 Layers Atop the Digital State

FIG 19.2 · THE OPTIMAL 8-LAYER NATIONAL AGENTIC TECHNOLOGY STACK
L8
GOVERNANCE SPINE (cross-cutting, all layers)
binding risk framework (NIST AI RMF / EU AI Act / SG Model Gov) · liability & accountability law · central agent authority · mandatory eval harness · oversight institution · workforce programme
L7
EXPERIENCE & MISSION
citizen super-assistant · civil-servant copilots · policy twin · proactive life-event services — bounded, human-gated
L6
ORCHESTRATION
stateful durable workflows · orchestrator-worker multi-agent · A2A inter-agency federation · human-approval nodes · recovery & retries
L5
AGENT RUNTIME
reason–act–verify loop · reflection / self-correction · planning · skill library · workflow-vs-agent discipline
L4
TRUST & CONTROL (cross-cutting)
agent identity · authenticated delegation (OAuth/OIDC+) · least-privilege action scopes · guardrails · tested kill-switch · observability + immutable logs · secure gateways
L3
TOOLS & INTEGRATION
MCP servers over every registry/system · tool & agent registry / discovery · computer-use legacy bridge (sandboxed)
L2
DATA, KNOWLEDGE & MEMORY
agent-ready authoritative registries · RAG · vector + graph stores · context engineering · governed tiered memory · consent layer · structured outputs
L1
COMPUTE & MODELS
governed frontier + sovereign / open-weight models · model gateway (optionality) · managed national inference · cost routing
▼   SITS ATOP THE DIGITAL STATE   ▼
DIGITAL-STATE FOUNDATION (folders 01–07) — universal broadband · eID · base registries · X-Road-style signed data exchange · sovereign cloud · zero-trust security · interoperability. The agentic stack fails without it.
The optimal national agentic architecture. A sovereign agent platform provisioned centrally for all of government, where every agent is a grounded, reflective reason–act–verify loop with governed memory, reaching the world through MCP and other agents through A2A, all under supervised autonomy and a binding national risk framework — sitting atop a healthy digital state. Magenta = load-bearing P0/P1 layers; L4 & L8 are cross-cutting. The win is not any single exotic pattern; it is having all the agentic P0–P1 components wired together onto a working digital foundation — which fewer than ~10 nations have done.
8layers · L1 compute → L8 governance
2cross-cutting · L4 Trust, L8 Governance
<10nations with the full stack (est.)
NATI · ENSI Report 2026The 10 Architectures
ENSI · National Agentic Technology Infrastructure46

PART 2 · WHY THIS IS THE OPTIMAL DESIGN NOW


Seven Load-Bearing Reasons

1 · THE PLATFORM PATTERN COMPOUNDS

Every credible leader — USAi, GenAI4EU, the Agentic State sovereign stack — converged on a shared platform because it lets a nation operationalise safety, identity, evaluation and cost-control once and inherit them everywhere. A thousand agency pilots cannot be governed; one platform can.

2 · GROUNDING MAKES AGENTS TRUSTWORTHY

The defining risk of LLM agents is confident error; the defining requirement of public administration is defensible decisions. RAG + authoritative registries + citations resolve the tension — the agent can always show its source. Ungrounded agents on official matters automate unaccountability.

3 · SUPERVISED AUTONOMY IS THE SURVIVABLE MODE

Unanimously across the governance sources, autonomous action on citizens must be bounded, gated, attributable, interruptible and overseen (EU AI Act, UK Playbook, NASCIO, OpenAI). This is not a brake on the architecture; it is a structural member of it — climb the maturity model, don't leap.

4 · IDENTITY & DELEGATION EXTEND THE DEEPEST PRINCIPLE

The digital state's load-bearing idea was a trusted identity at the root of every transaction. The agentic state extends it: every agent has a verifiable identity (arXiv 2401.13138) and acts only under scoped, revocable, authenticated delegation (arXiv 2501.09674) — letting an agent act on behalf of a citizen without dissolving accountability.

5 · MCP + A2A ARE FEDERATION REBORN

The single best digital archetype was X-Road's federation — autonomous nodes interoperating through a standard, no central monolith. MCP (agent-to-tool) and A2A (agent-to-agent) reproduce exactly that shape: integrate through open standards, preserve agency ownership, avoid a central black box.

6 · MEMORY + REFLECTION MAKE DURABLE WORKERS

Stateless, single-shot agents are toys. Tiered memory (MemGPT, Generative Agents) and self-correction (Reflexion, verify stages) make agents reliable across long-horizon casework — provided memory is governed as regulated personal data. Architecture and governance are inseparable.

7 · YOU EARN THE INTELLIGENCE LAYER

The Algorithmic State Architecture models intelligence as the capstone of DPI → Data → Interoperability → Intelligence. L1–L7 are useless without the digital state beneath them, and L7 (mission) is dangerous without L1–L6. The order is the strategy.

THE SPINE

The agentic state is the digital state's canon — identity, registries, federation, interoperability, governance, oversight — extended to a worker that reasons and acts. The win is having the agentic P0–P1 layer wired onto a healthy digital foundation — which fewer than ~10 nations have done.

The win is not any single exotic pattern; it is having all of the agentic P0–P1 components wired together onto a working digital foundation — which fewer than ~10 nations have done, and perhaps two or three have done well.

Synthesis C, Thesis of the Optimal National Agentic Stack
NATI · ENSI Report 2026The 10 Architectures
ENSI · National Agentic Technology Infrastructure47

PART 2 · THE BUILD SEQUENCE


What an Optimal Nation Does, In Order

The reference design is also a sequence. Each step depends on the one beneath it; the discipline is to refuse to ship autonomy before the trust spine exists, and to roll out mission last and continuously.

  1. Pre-requisite — fix the digital state. Ensure eID, agent-ready authoritative registries, X-Road-style data exchange, sovereign cloud and zero-trust security are healthy (folders 01–07). Without this, stop here — no amount of agent procurement substitutes for the foundation.
  2. L1–L2 foundation. Stand up governed model access behind a gateway, managed national inference, the RAG + vector/graph data layer, context engineering, and governed tiered memory.
  3. L3–L4 trust spine. Expose systems as MCP servers behind secure gateways; build agent identity, authenticated delegation, least-privilege scopes, observability/immutable logging and a tested kill-switch. Nothing autonomous ships before L4 exists.
  4. L5–L6 runtime. Deploy the reason–act–verify loop with reflection and planning; add stateful orchestration, orchestrator-worker multi-agent, A2A federation, and human-approval nodes.
  5. L8 governance, in parallel from day one. Adopt a binding risk framework, stand up the eval harness and oversight institution, legislate liability, and run the workforce programme.
  6. L7 mission, last and continuously. Roll out citizen super-assistants, civil-servant copilots, policy twins and proactive services — starting where the time analysis says the burden is highest and the risk lowest, climbing the NASCIO maturity model as reliability is proven.
8+US states run agentic tools in production (NASCIO 2026)
1platform governs what 1,000 pilots cannot
L4the gate before any autonomy ships
THE SPINE, RESTATED

The optimal national agentic architecture, in one sentence, is the same sentence as the digital one — extended by a single, world-changing clause: a federated, API-first, building-block platform with a trusted identity at its root and a signed exchange layer instead of a central database — now staffed by governed, grounded, supervised agents that reason and act on top of it. No nation has the full stack. The frontier is a composite — the US's platform, Estonia's digital foundation and framework, the UK's assurance discipline, Singapore's governance toolkit, the EU's binding risk law, the UAE's mission ambition — and the decisive fact is that the agentic race is being run on the foundations laid by the digital one.

NATI · ENSI Report 2026The 10 Architectures
ENSI · National Agentic Technology Infrastructure48
MOVEMENT D · §22

Who's Closest — The Agentic-State Scorecard

No nation has the full optimal stack. The frontier is a composite — the US's platform, Estonia's digital foundation, the UK's assurance discipline, Singapore's governance toolkit, the EU's binding risk law, the UAE's mission ambition. The seven profiles that follow rank each on the same axis: how much of the agentic P0–P1 layer is wired onto a healthy digital state — the thing fewer than ~10 nations have done.

CASE STUDY 01 · UNITED STATES (FEDERAL)

USAi & the sovereign agent platform


The United States is closest on the single highest-leverage move in this collection — the shared, government-wide agent platform. GSA's USAi gives every agency a unified chat, API and evaluation console over multiple leading models, free at the point of use; OneGov buys "government as one customer" with agentic deals already struck; and the FDA has taken agents from pilot to an agency-wide premarket-review rollout.

The platform pattern is the US's decisive advantage: where weaker states accumulate a thousand ungoverned pilots, Washington operationalises model access, cost-routing and an eval console once and lets agencies inherit them. The AI Action Plan ("Winning the Race") supplies the compute-and-infrastructure pillar beneath it. The gap is governance coherence: a fast, deliberately deregulatory rollout risks outrunning the supervised-autonomy discipline the rest of this report treats as load-bearing.

1
government-wide USAi platform (chat + API + eval console, multi-model)
GSA / FEDERAL AI LANDSCAPE 2025
90+
actions in the AI Action Plan "Winning the Race"
WHITE HOUSE 2025
Strongest archetype

#1 Sovereign Agent Platform USAi is the live exemplar of government-as-a-platform, agentic edition — the agentic successor to Singapore's SGTS and the UK's GDS.

Mini-timeline
  • 2025 — AI Action Plan published (3 pillars, 90+ actions)
  • 2025 — GSA launches USAi government-wide
  • 2025 — OneGov agentic procurement deals (ServiceNow, Meta)
  • 2026 — FDA premarket-review agents at agency scale
Agentic stack coverage — indicative readiness (est., 0–100)
L1 Compute/models88
L2 Data & memory78
L3 Tools / MCP65
L4 Trust & control60
L5 Agent runtime62
L6 Orchestration90
L7 Experience/mission75
L8 Governance spine58
MAIN GAP — COHERENT GOVERNANCE AT SPEED

Platform maturity is racing ahead of a unifying supervised-autonomy and oversight spine. The strategic question is whether the US can keep grounding, eval harnessing and human-gating in step with a deregulatory rollout — or automate unaccountability at machine speed. The platform is there; the governing wall must keep pace.

NATI · ENSI Report 2026Evidence & Back Matter
ENSI · National Agentic Technology Infrastructure50

CASE STUDY 02 · ESTONIA

The deepest foundation — and the framework that named the era


Estonia did not start the agentic race; it defined it. The Agentic State vision paper (Ilves/Kilian) is the first framework for the agentic transformation of core government functions, and it could only have come from the nation with the deepest digital floor to build on: X-Road federation, population-scale eID, clean base registries and KSI tamper-evident logging.

This is the inheritance thesis made flesh. Estonia begins the agentic race with half the P0 layer already laid — a signed data-exchange bus, a trusted identity at the root of every transaction, and a RIHA registry of trusted services that is the natural ancestor of an agent registry. Its strongest archetype is therefore federation reborn at the agent layer: autonomous nodes interoperating through a standard, no central monolith. The gap is execution scale — moving from a world-leading framework to a production agent platform at USAi's reach.

1st
framework for agentic transformation of core gov functions (Agentic State)
AGENTIC STATE VISION PAPER 2025
~99%
e-services digital; eID + base registries agent-ready (est.)
e-ESTONIA GUIDE (est.)
Strongest archetype

#5 Federated Inter-Agent Mesh plus the capstone intelligence layer — the agentic reincarnation of the #1 digital archetype, X-Road federation.

Mini-timeline
  • 2001 — X-Road data-exchange layer goes live
  • 2007– — KSI tamper-evident state logging
  • 2025 — Agentic State vision paper sets the agenda
  • 2026 — framework → production platform (in progress)
MAIN GAP — FRAMEWORK TO PRODUCTION

Estonia has the design and the foundation; what it must now demonstrate is the scaled, governed production platform on top of them. The enablers the Agentic State paper itself names — agent governance, a sovereign tech stack, cyber and people — are the build-out still ahead. The foundation is unmatched; the capstone is half-built.

Agentic stack coverage — indicative readiness (est., 0–100)
L1 Compute/models92
L2 Data & memory85
L3 Tools / MCP70
L4 Trust & control72
L5 Agent runtime88
L6 Orchestration66
L7 Experience/mission68
L8 Governance spine72
NATI · ENSI Report 2026Evidence & Back Matter
ENSI · National Agentic Technology Infrastructure51

CASE STUDY 03 · UNITED KINGDOM

Assurance discipline and an evidence-led target


The UK leads on the dimension that makes autonomous action survivable: supervised autonomy and assurance. The AI Playbook for the UK Government sets ten principles plus procurement and assurance guidance for safe public-sector AI, and the Alan Turing Institute supplies the evidence base that tells the state where to point agents first.

Turing's Mapping the Potential quantified that ~41% of public-sector working time in Great Britain is spent on activities supportable by generative AI (49% in education) — exactly the "start where the time is, not where the hype is" discipline. Combined with the Playbook's human-oversight and assurance requirements, the UK is the clearest exemplar of the human-in-the-loop, earn-your-autonomy model. The gap is a unifying national agent platform at USAi scale — the UK has the governance and the targeting, but not yet the single shared substrate the leaders converge on.

10
principles in the AI Playbook + procurement & assurance guidance
UK GDS / CABINET OFFICE 2025
GOV.UK
One Login + Service Standard — the digital floor beneath the playbook
UK GDS (LINK-ONLY)
Strongest archetype

#4 Supervised Autonomy The only pattern that appears in every credible governance source — and the one the UK has institutionalised earliest through the Playbook + assurance.

Mini-timeline
  • 2025 — AI Playbook (10 principles) published
  • 2025 — Turing "Mapping the Potential" evidence base
  • 2025 — GOV.UK One Login scales the identity floor
  • 2026 — national agent platform (the missing piece)
Agentic stack coverage — indicative readiness (est., 0–100)
L1 Compute/models70
L2 Data & memory68
L3 Tools / MCP60
L4 Trust & control72
L5 Agent runtime64
L6 Orchestration55
L7 Experience/mission66
L8 Governance spine88
MAIN GAP — A UNIFYING NATIONAL PLATFORM

The UK has the assurance discipline and the evidence-led targeting but lacks a single shared agent platform at USAi's reach. Governance without a platform governs nothing at scale; the next move is to wire the ten principles into one inherited substrate every department builds on.

NATI · ENSI Report 2026Evidence & Back Matter
ENSI · National Agentic Technology Infrastructure52

CASE STUDY 04 · SINGAPORE

NAIS 2.0, the nine-dimension framework and AI Verify


Singapore pairs the deepest public-service AI embedding with the world's most operational governance toolkit. National AI Strategy 2.0 (three systems / ten enablers / fifteen actions) drives AI deep into public service, while the Model AI Governance Framework for GenAI sets nine governance dimensions, paired with the AI Verify testing toolkit that turns governance from a PDF into a test you can run.

This is the platform-plus-governance archetype: SGTS already gave Singapore a government-as-a-platform base layer and reusable whole-of-government APIs, and Singpass gives ~97% identity coverage. The combination of NAIS 2.0's embedding and AI Verify's testability makes Singapore the model for measurable, conformant agentic AI. The gap is the move from copilots — AI that assists — to bounded autonomy — AI that acts — under those same controls.

9
dimensions in the Model AI Governance Framework for GenAI
IMDA / AI VERIFY 2024
~97%
Singpass identity coverage — the agent-delegation foundation
GOVTECH SG
Strongest archetype

#1 Platform + #8 Reflective / Eval AI Verify operationalises the eval-harness discipline (P49) — testability is governance made real.

Mini-timeline
  • 2021 — SGTS base + services layer in production
  • 2023 — NAIS 2.0 published
  • 2024 — Model AI Gov Framework + AI Verify toolkit
  • 2026 — copilots → bounded autonomy (next)
MAIN GAP — FROM COPILOTS TO BOUNDED AUTONOMY

Singapore's governance and embedding are exemplary, but its agents still mostly assist rather than act. The next move is to climb the maturity model into supervised, gated autonomy — applying the nine dimensions and AI Verify to agents that take consequential actions, not only to assistants that draft them.

Agentic stack coverage — indicative readiness (est., 0–100)
L1 Compute/models85
L2 Data & memory80
L3 Tools / MCP72
L4 Trust & control70
L5 Agent runtime74
L6 Orchestration78
L7 Experience/mission80
L8 Governance spine85
NATI · ENSI Report 2026Evidence & Back Matter
ENSI · National Agentic Technology Infrastructure53

CASE STUDY 05 · UNITED ARAB EMIRATES

Mission ambition and a Minister of AI


The UAE leads on the experience-and-mission layer — the vision of agents as front-line public servants. The National Strategy for AI 2031 frames agents as the "workhorses that convert policy into daily service," and the world's first Minister of State for AI supplies the central, top-down authority the platform pattern needs.

The UAE's strength is mission clarity and institutional will: a single accountable office, a strong platform drive, and a concrete agents-in-service vision spanning visas, logistics and citizen services, all on the UAE PASS eID, signature and SSO foundation. This is the "treat agents as a new factor of production" principle taken seriously at the top of government. The gap is the governance counterweight — open eval depth and transparent oversight to match the mission ambition, so autonomy is earned rather than simply mandated.

1st
nation with a Minister of State for AI — central agentic authority
UAE AI OFFICE (LINK-ONLY)
UAE PASS
eID + signature + SSO — the delegation foundation
TDRA DIGITAL GOV ENABLERS
Strongest archetype

#7 Mission / Experience Citizen super-assistants and proactive life-event services (visas, logistics) — agents at the front line of service delivery.

Mini-timeline
  • 2017 — world's first Minister of State for AI
  • 2018 — National AI Strategy 2031 launched
  • 2023 — UAE PASS / GovSign enablers scaled
  • 2026 — agents-in-service across visas & logistics
Agentic stack coverage — indicative readiness (est., 0–100)
L1 Compute/models82
L2 Data & memory72
L3 Tools / MCP60
L4 Trust & control56
L5 Agent runtime68
L6 Orchestration64
L7 Experience/mission85
L8 Governance spine58
MAIN GAP — OPEN GOVERNANCE & EVAL DEPTH

The mission ambition outpaces the published governance. To make top-down autonomy survivable, the UAE needs open eval harnesses, transparent oversight institutions and a binding risk framework on par with its delivery drive — so that agents that act on citizens are demonstrably grounded, gated and accountable.

NATI · ENSI Report 2026Evidence & Back Matter
ENSI · National Agentic Technology Infrastructure54

CASE STUDY 06 · EUROPEAN UNION (AS A BLOC)

Apply AI, GenAI4EU and the world's binding risk law


The EU brings the one thing no other actor has: a binding risk framework. The EU AI Act (Regulation 2024/1689) is the world's first horizontal, enforceable AI law — risk tiers, high-risk requirements, mandatory human oversight — and the Apply AI / GenAI4EU programmes give the bloc a sovereign-platform play with explicit agentic-AI support for public administration.

The EU's strongest archetypes are therefore federation and governance: the AI Act as the compliance perimeter every member state must build inside, plus A2A/AGNTCY interoperability momentum that mirrors the bloc's deep heritage in cross-border interoperability (EIF, eIDAS, Once-Only). The gap is speed — turning binding frameworks and funded programmes into deployed, production platforms before the frontier moves on. The EU writes the rules of the road faster than it builds the road.

1st
binding, horizontal AI law in the world (risk tiers + human oversight)
EU AI ACT 2024/1689
A2A
AGNTCY / Agent2Agent interoperability momentum — federation reborn
LINUX FOUNDATION 2025
Strongest archetype

#5 Federation + #4 Governance The AI Act is the binding spine (P50); A2A/AGNTCY is X-Road federation reborn at the agent layer.

Mini-timeline
  • 2017 — EIF; 2014 eIDAS interoperability heritage
  • 2024 — EU AI Act enters into force
  • 2025 — Apply AI / GenAI4EU launched
  • 2026 — frameworks → deployed platforms (the race)
Agentic stack coverage — indicative readiness (est., 0–100)
L1 Compute/models72
L2 Data & memory78
L3 Tools / MCP66
L4 Trust & control80
L5 Agent runtime64
L6 Orchestration70
L7 Experience/mission62
L8 Governance spine92
MAIN GAP — SPEED OF DEPLOYMENT

The EU's binding law and funded programmes are world-leading; its risk is turning them into production platforms slower than the frontier moves. The challenge is institutional velocity — shipping the deployed agentic infrastructure as fast as it ships the regulation.

NATI · ENSI Report 2026Evidence & Back Matter
ENSI · National Agentic Technology Infrastructure55

CASE STUDY 07 · LEADING US STATES

Agents in production — the maturity-model vanguard


The most concrete agentic-production evidence in this entire collection comes not from a national government but from the US states. NASCIO's 2026 report, Beyond Generation, documents 8+ states already running agentic tools in production — and crucially, they reached them by climbing a five-phase maturity model, not leaping.

That maturity model — assistive GenAI → context-aware GenAI → task-level automation → stateful multi-step workflows → adaptive/proactive agents — is the emblematic sequencing instrument of the whole field, and the states are its proof of concept: supervised autonomy earned level by level, with oversight at each prior level before more autonomy is granted. The strongest archetype is supervised autonomy. The gap is the shared platform and the identity/delegation depth that a national actor like USAi can provide but a single state cannot easily build alone.

8+
US states running agentic tools in production (hard signal)
NASCIO 2026
acts
the shift NASCIO names — "AI that assists" → "AI that acts"
NASCIO 2026
FIG D-1
P1 Assistive GenAIwide
P2 Context-aware GenAIcommon
P3 Task-level automationgrowing
P4 Stateful workflows8+ states
P5 Adaptive / proactivefrontier
The NASCIO maturity climb. Leading US states reach production agents (P4) by ascending each phase — not by leaping to autonomy. Bar widths are illustrative of relative adoption breadth across states (est.); the 8+ in-production figure is the hard NASCIO signal.
Agentic stack coverage — indicative readiness (est., 0–100)
L1 Compute/models66
L2 Data & memory62
L3 Tools / MCP55
L4 Trust & control60
L5 Agent runtime56
L6 Orchestration60
L7 Experience/mission80
L8 Governance spine64
MAIN GAP — SHARED PLATFORM + DELEGATION DEPTH

The states have the discipline; what they lack is a shared agent platform and the identity/authenticated-delegation depth a national actor provides. Fifty separate climbs of the same maturity model is the federal case for a USAi-class substrate underneath them.

NATI · ENSI Report 2026Evidence & Back Matter
ENSI · National Agentic Technology Infrastructure56

MOVEMENT D · §23 — CROSS-CUTTING

23 What the leaders share — and where states fail


Read across the seven profiles and the pattern is unmistakable. No nation has the full optimal stack, but every leader converged on the same small set of moves — and every failure traces to the same small set of omissions. The order is the strategy.

What the leaders share

A shared platform, not piecemeal procurement. USAi, GenAI4EU and the Agentic State's sovereign stack all converged on one substrate because safety, identity, evaluation and cost control are operationalised once and inherited everywhere. A thousand agency pilots cannot be governed; one platform can.

A working digital foundation underneath. Estonia's X-Road, Singapore's SGTS, the UK's One Login, the UAE's UAE PASS — every leader is building agents on a healthy identity-and- registry floor. The agentic race is being run on the foundations laid by the digital race.

Supervised autonomy, earned by maturity. The UK Playbook, NASCIO's five phases, the EU AI Act and OpenAI's practices all insist autonomy be bounded, gated, attributable and interruptible — and climbed, not leapt. This is a structural member of the architecture, not a brake on it.

Grounding and an eval harness. Singapore's AI Verify and the τ-bench/AgentBench/GAIA discipline make correctness measurable. Leaders deploy agents that can show their source and pass a documented suite before — and during — production.

THE FOUR FAILURE MODES — HOW STATES AUTOMATE DYSFUNCTION

1 · Buying agents without the foundation. Procuring autonomous tools onto broken registries, missing eID and un-API'd systems. The agent faithfully retrieves and amplifies garbage at machine speed; the result is an expensive demo, not infrastructure.
2 · Ungoverned pilots. A thousand agency experiments with no central platform — no shared guardrails, no shared observability, every ministry re-failing at safety alone.
3 · No eval harness. Deploying without measuring capability and reliability under repetition. τ-bench's pass^k finding — that agents fail on repeated-trial consistency — is the single most important deployment caution in this collection, and it is invisible without a harness.
4 · No kill-switch, no human gate. Autonomous action on citizens with no tested emergency stop, no rollback and no meaningful approval gate. An autonomy you cannot halt is an autonomy you cannot deploy.

The decisive strategic fact is that the agentic race is being run on the foundations laid by the digital race — the nations that built X-Road-class interoperability, population-scale identity and clean registries start with half the agentic P0 stack already in place.

Synthesis C, Ten Agentic Architectures & the Optimal National Stack
NATI · ENSI Report 2026Evidence & Back Matter
ENSI · National Agentic Technology Infrastructure57

MOVEMENT D · §23 — THE BUILD-ORDER INSIGHT

24 The recommendations matrix


The build-order insight is the whole strategy: you earn the intelligence layer, you do not skip to it. The matrix below sequences the synthesis build order into ten concrete actions — fix the digital state, stand up L1–L2, build the L3–L4 trust spine before anything autonomous ships, then runtime, governance in parallel, and mission last and continuously.

ActionOwnerBy whenMeasured how
0 · Fix the digital state — eID, agent-ready registries, X-Road-style exchange, sovereign cloud, zero-trustCentral digital agencyPre-requisiteRegistry quality/lineage score; eID coverage %; data-exchange uptime
1 · Stand up the national agent platform — shared model access behind a gateway, tools, memory, observabilityCentral agent authority0–12 moAgencies onboarded; marginal cost-per-agency → 0
2 · Model gateway + managed inference — multi-model optionality, cost routing, sovereign/open-weight fallbackPlatform team0–12 moModels routable; cost-per-resolution; sovereign inference path live
3 · RAG + registry grounding layer — vector + graph stores, citations, structured outputsData office6–18 mo% of agent claims grounded & cited; hallucination defect rate
4 · Expose systems as MCP servers behind policy-enforcing gateways; tool & agent registryPlatform + agencies6–24 moRegistries exposed as governed tools; gateway-mediated calls %
5 · Agent identity + authenticated delegation — scoped, revocable, logged; nothing autonomous ships before thisIdentity authority6–18 moAgents with verifiable identity %; delegation scopes audited
6 · Observability, kill-switch & human gates — immutable logs, tested emergency stop, approval nodesCentral agent authority6–18 moKill-switch drill pass; % consequential actions gated; log coverage
7 · Eval harness, mandatory — capability + pass^k reliability + safety + policy-adherence; re-qualify on every model changeOversight institution0–18 mo, continuous% production agents passing documented suite; re-qual cadence
8 · Binding risk framework + liability law — NIST AI RMF / EU AI Act / SG Model Gov; named responsible human per actionLegislature + central authority0–24 mo, parallelEvery system mapped to a risk tier; accountability chain legislated
9 · Roll out mission services — citizen super-assistant, civil-servant copilots, proactive life-events — highest-burden, lowest-risk firstLine agencies12 mo+, continuousCitizen outcomes resolved; civil-servant hours returned; maturity-phase
THE BUILD-ORDER INSIGHT

Rows 0–8 are the trust spine; row 9 is the only one citizens see — and it ships last, on top of everything above it. The Algorithmic State Architecture models intelligence as the capstone of DPI → Data → Interoperability → Intelligence. L1–L7 are useless without the digital state beneath them, and L7 (mission) is dangerous without L1–L6. The order is the strategy.

NATI · ENSI Report 2026Evidence & Back Matter
ENSI · National Agentic Technology Infrastructure58

MOVEMENT E · §24 — RISKS & COUNTER-ARGUMENTS

25 Could the thesis be wrong?


A confident report owes its reader the strongest case against itself. The thesis — that nations should wire the agentic P0–P1 layer onto a healthy digital foundation now — faces four serious objections, each grounded in this collection's own evidence. None overturns the spine; each sharpens it.

OBJECTION 1 · AGENT RELIABILITY

The pass^k problem. τ-bench's central finding is that agents which pass a task once often fail it on repeated trials — consistency, not peak capability, is the binding constraint. GAIA reports models at ~15% on real-world assistance where humans score ~92%. A state that deploys agents on consequential casework before reliability is proven is automating an unaccountable error rate. Rebuttal: this is precisely why the eval harness (P49) and supervised autonomy (P40) are foundational, not optional — the thesis requires measuring pass^k and gating on it, which is the opposite of naïve deployment.

OBJECTION 2 · HYPE CYCLE

The standards are two years old. A2A was donated to the Linux Foundation only in 2025; MCP is young and evolving; the agentic archetypes are mostly proven in the private sector and in pilots, not as governed national infrastructure. Building national plumbing on moving standards risks expensive churn. Rebuttal: the load-bearing investments — clean registries, eID, delegation, observability, an eval harness — are standard-agnostic. They pay off regardless of which agent protocol wins, which is why the build sequence front-loads them and treats the protocols as the replaceable top layer.

OBJECTION 3 · SAFETY INCIDENTS

Agent memory is a surveillance apparatus in waiting. Long-horizon agents that remember citizens, and computer-use agents with general UI control, are the two highest-risk configurations in this report. A single ungoverned incident could set public trust back a decade. Rebuttal: the architecture is inseparable from its governance — governed tiered memory as regulated personal data (P35), least-privilege scopes, kill-switch and immutable logs are not add-ons but structural members. The thesis bans the dangerous configuration, it does not enable it.

OBJECTION 4 · THE OECD "LIMITED PRECEDENT" CAUTION

The OECD's survey of 200 government AI use cases found the public sector has "limited precedent for autonomous software in government" while the private sector races ahead. This is the steelman's strongest card: government has never run software that acts on citizens at scale, and caution is warranted. But the OECD frames it as both a warning and an opportunity — the precedent gap is exactly why building the governed foundation now, deliberately and in sequence, is the responsible path rather than waiting for the private-sector pattern to arrive ungoverned.

NATI · ENSI Report 2026Evidence & Back Matter
ENSI · National Agentic Technology Infrastructure59

MOVEMENT E · §24 — OUTLOOK

26 Base, bull and bear to 2030


Where does national agentic adoption land by ~2030? The honest answer is a fan of scenarios, anchored to the one hard signal we have — fewer than ~10 nations have wired the agentic P0–P1 layer onto a healthy digital foundation today.

FIG E-1
Today (2026) · foundation + P0–P1 wired<10
2030 · BEAR — frameworks stall, pilots churn~15
2030 · BASE — leaders scale, fast-followers arrive~30
2030 · BULL — platform pattern + open standards diffuse~50
Nations with a governed agentic P0–P1 layer on a healthy digital state. Reasoned scenario estimates (est.) from the <10-nation 2026 baseline. Spread is governance velocity, not model capability.
ScenarioWhat drives itWhat it looks like in 2030
BEARFrameworks stall as PDFs; pilots churn; a safety incident freezes trust; standards fragment~15 nations; agents stuck at assistive copilots; autonomy politically radioactive
BASEThe platform pattern proves out; leaders scale to production; fast-followers inherit USAi/AI-Act-class blueprints~30 nations with governed P0–P1; supervised autonomy normal in high-volume, low-risk functions
BULLMCP/A2A standards diffuse; shared platforms + open eval harnesses become reusable public goods~50 nations; the agentic capstone becomes the new digital-government baseline, as e-services did after 2010
<10nations with the layer wired today (est.)
198economies in the World Bank GTMI
~38digital governments OECD-benchmarked
8+US states already in production (NASCIO)
THE THROUGH-LINE OF EVERY SCENARIO

The variable that separates bear from bull is not model capability — it is governance velocity and whether nations build the foundation first. The states that built X-Road-class interoperability and population-scale identity start every scenario with half the agentic P0 stack in place; the rest find no amount of agent procurement substitutes for it.

NATI · ENSI Report 2026Evidence & Back Matter
ENSI · National Agentic Technology Infrastructure60

MOVEMENT E · §25 — METHODOLOGY

27 How this evidence base was built


This report rests on a curated, verified corpus assembled with a single discipline: every URL was checked to resolve, only real multi-page PDFs were downloaded, and living specifications were recorded as links, not faked as downloads — an evidence base you can audit, not a bibliography you must trust.

12thematic folders (7 digital + 5 agentic)
12parallel research teams (7 + 5), one per folder
142real PDFs downloaded & checked (96 + 46)
~174sources catalogued in total (111 + 63)

The four-part discipline

Verify every URL. Each source was found via live web search and its URL verified to resolve before it entered the catalogue. Nothing was cited from memory or assumed to exist.

Only real PDFs. Where a free PDF existed, it was downloaded and checked with file / pdfinfo to confirm it was a real, multi-page document. 142 sources passed this bar (96 digital + 46 agentic).

Link-only for living specs. Canonical specs published as living HTML or Git (X-Road ARC-G, GovStack, MOSIP, MCP, A2A, AGNTCY) and bot-blocked pages (OECD, UAE, EU, GSA) were recorded as LINK-ONLY with a working URL — never fabricated as downloads.

Grounding & est. discipline. Every number traces to the folder-04 indices or a named source. Agentic infrastructure is ~2 years old with no ITU-grade census, so most adoption figures carry an explicit (est.) label; only hard signals — 8+ US states (NASCIO), the FDA rollout, USAi/GenAI4EU — are cited as fact.

100%
of URLs verified to resolve before cataloguing
INTEGRITY NOTE
0
fabricated downloads — living specs marked LINK-ONLY, never faked
METHODOLOGY
WHY THE DISCIPLINE MATTERS FOR AN AGENTIC REPORT

A report arguing that states must ground their agents in verifiable, cited, authoritative sources cannot itself rest on unverified claims. The grounding rule we recommend for national agents — show your source, label your estimates — is the same rule this evidence base was built under. Method is the argument.

NATI · ENSI Report 2026Evidence & Back Matter
ENSI · National Agentic Technology Infrastructure61

MOVEMENT E · §26 — THE EVIDENCE BASE

28 The corpus — 12 folders, ~174 sources


The full collection is twelve thematic folders — the seven that document the digital state and the five that document its agentic capstone. The counts below are the catalogued totals; together they form the spine of every claim in this report.

142
real PDFs downloaded & verified (96 digital + 46 agentic)
00_README + AGENTIC
~45
sources on real national architectures (~40% of the digital base)
00_README
#FolderScopeSources
01Architectures · EuropeEstonia X-Road, Finland, Denmark, Norway, NL, Germany — real stacks16
02Architectures · AsiaSingapore, India Stack, S. Korea, Japan, China, UAE — real stacks16
03EU/UK/US frameworksEIF, eIDAS/EUDI, NIST, FedRAMP, FEAF, GAIA-X, Zero Trust15
04Indices & frameworksITU, UN, World Bank, OECD, GSMA, NRI — the measurement layer15
05Digital Public InfrastructureUNDP/G20, GovStack, MOSIP, Mojaloop, X-Road, DEPA15
06Technical architectureBroadband, 5G/6G, Open-RAN, IXP, RPKI, DNS, edge, sovereign cloud, ZT18
07Emerging-nation referenceSmart Africa, Diia, PIX, Rwanda, ASA, GEA reference models16
08Agentic strategies · nationalAgentic State, NASCIO, UK Playbook, NAIS 2.0, USAi, AI Action Plan, OECD, TBI, Turing, UAE, EU13
09Agent architectures & patternsReAct, Reflexion, Toolformer, ToT, CoT, surveys, Generative Agents, Voyager13
10Requirements & foundationsNIST AI RMF, EU AI Act, Visibility, OpenAI practices, AgentBench/GAIA/τ-bench, SG Model Gov11
11Agent data & stateRAG, MemGPT, Generative Agents, GraphRAG, KG+LLM, context engineering, OECD govt-data10
12Agent tools & integrationMCP, A2A, Gorilla, ToolLLM, API-Bank, WebArena, computer-use, AutoGen, LangGraph, AGNTCY12
7 + 5folders: digital state + agentic capstone
~45sources on real agent architectures (folders 08–12)
~17canonical specs kept LINK-ONLY, never faked
NATI · ENSI Report 2026Evidence & Back Matter
ENSI · National Agentic Technology Infrastructure62

MOVEMENT E · §26 — THE EVIDENCE BASE

29 The marquee sources


From ~174 catalogued sources, these ~14 are the load-bearing references — the documents the principles, components and architecture rankings most directly rest on. Each is a real, verified source or a canonical LINK-ONLY specification.

Governance, risk & identity

National strategies & adoption

Agent architecture & tools

Digital-state foundation

142real PDFs
~174sources total
12thematic folders
FROM CATALOGUE TO CANON

These marquee sources are why the report can make its central claim with confidence: the agentic state is the digital state's canon — identity, registries, federation, interoperability, governance, oversight — extended to a worker that reasons and acts. Every one of those words traces to a verified document above.

NATI · ENSI Report 2026Evidence & Back Matter
ENSI · National Agentic Technology Infrastructure63
ENSI · NATIONAL AGENTIC TECHNOLOGY INFRASTRUCTURE

The order is the strategy

The agentic state is not a leap past the digital state — it is the digital state's canon, identity, registries, federation, interoperability, governance and oversight, extended to a worker that reasons and acts. An agent is only as good as the identity it can verify, the registry it can trust, the tool it can call, and the human who can stop it. The win is not any single exotic pattern; it is having the agentic P0–P1 layer wired onto a healthy digital foundation — which fewer than ~10 nations have done, and perhaps two or three have done well. The nations that built X-Road-class interoperability, population-scale identity and clean registries start the agentic race with half the foundation already laid; the rest will find that no amount of agent procurement substitutes for it. Build the foundation. Earn the capstone. Govern the worker that acts.
50NATI principles
50agentic components
10architecture archetypes
~174verified sources
ENSI · European National Strategy & Intelligence Report · National Agentic Technology Infrastructure Cite as · ENSI, NATI: The Agentic State (June 2026) Contact · intelligencestrategy.org