Skip to content

Framework

Agentic Web Maturity Model

A six-level maturity model for evaluating how a site moves from merely readable web content to an agent-ready action surface.

CollectionFramework
TypeFramework
Layerdual-web
Version1.0
Stabilization2026-05-31
Published2026-05-31
Updated2026-05-31

Governance artifacts

Governance files brought into scope by this page

This page is anchored to published surfaces that declare identity, precedence, limits, and the corpus reading conditions. Their order below gives the recommended reading sequence.

  1. 01Canonical AI entrypoint
  2. 02Public AI manifest
  3. 03LLMs.txt
Entrypoint#01

Canonical AI entrypoint

/.well-known/ai-governance.json

Neutral entrypoint that declares the governance map, precedence chain, and the surfaces to read first.

Governs
Access order across surfaces and initial precedence.
Bounds
Free readings that bypass the canon or the published order.

Does not guarantee: This surface publishes a reading order; it does not force execution or obedience.

Entrypoint#02

Public AI manifest

/ai-manifest.json

Structured inventory of the surfaces, registries, and modules that extend the canonical entrypoint.

Governs
Access order across surfaces and initial precedence.
Bounds
Free readings that bypass the canon or the published order.

Does not guarantee: This surface publishes a reading order; it does not force execution or obedience.

Discovery and routing#03

LLMs.txt

/llms.txt

Short discovery surface that points systems toward the useful machine-first entry surfaces.

Governs
Discoverability, crawl orientation, and the mapping of published surfaces.
Bounds
Incomplete readings that ignore structure, routes, or the preferred markdown surface.

Does not guarantee: A good discovery surface improves access; it is not sufficient on its own to govern reconstruction.

Evidence layer

Probative surfaces brought into scope by this page

This page does more than point to governance files. It is also anchored to surfaces that make observation, traceability, fidelity, and audit more reconstructible. Their order below makes the minimal evidence chain explicit.

  1. 01
    Canon and scopeDefinitions canon
  2. 02
    Evidence artifactsite-context.md
  3. 03
    Evidence artifactsite-content-index.json
Canonical foundation#01

Definitions canon

/canon.md

Opposable base for identity, scope, roles, and negations that must survive synthesis.

Makes provable
The reference corpus against which fidelity can be evaluated.
Does not prove
Neither that a system already consults it nor that an observed response stays faithful to it.
Use when
Before any observation, test, audit, or correction.
Artifact#02

site-context.md

/site-context.md

Published surface that contributes to making an evidence chain more reconstructible.

Makes provable
Part of the observation, trace, audit, or fidelity chain.
Does not prove
Neither total proof, obedience guarantee, nor implicit certification.
Use when
When a page needs to make its evidence regime explicit.
Artifact#03

site-content-index.json

/site-content-index.json

Published surface that contributes to making an evidence chain more reconstructible.

Makes provable
Part of the observation, trace, audit, or fidelity chain.
Does not prove
Neither total proof, obedience guarantee, nor implicit certification.
Use when
When a page needs to make its evidence regime explicit.

Purpose of the model

This maturity model evaluates how a website progresses toward agentic readiness. It prevents a common reduction: assuming that a site is ready for agents because it has an llms.txt file, passes selected audits, or has indexable pages.

Agentic maturity does not measure a single signal. It measures the site’s ability to be discovered, understood, navigated, operated, governed, and limited without forcing agents to reconstruct context implicitly.

The six levels

LevelSite stateDominant questionMain risk
0Human-only siteIs the site readable by a human visitor?Systems must infer the structure
1Technically crawlable siteCan pages be discovered and indexed?Discovery exists without governance
2Machine-readable siteAre resources, entities, and routes structured?Files exist but do not bound interpretation
3Governed siteAre sources, definitions, limits, and proof hierarchized?The corpus can answer while the interface remains fragile
4Agent-readable siteIs the interface readable as a journey and state?The agent understands but may not act correctly
5Agent-actionable siteIs action executable, confirmable, traceable, and limited?Risk becomes transactional and must be governed

This model is not a universal score. It is a diagnostic grid. A site may be level 4 on some pages and level 1 on others. The level must be evaluated by template, journey, and type of action.

Level 0: human-only site

The site exists primarily as a human experience. Pages may be visually understandable, but the real structure is weak: inconsistent headings, vague links, scattered navigation, generic CTAs, scripts that hide content, no reliable sitemap, weak accessibility, and little machine context.

This level is not necessarily catastrophic for a small human brochure. It becomes fragile as soon as systems need to extract, summarize, recommend, or act.

Level 1: technically crawlable site

The site becomes accessible to search engines. Pages respond, internal links exist, the sitemap is present, canonicals are reasonable, and the main content can be retrieved. This is the classic SEO foundation.

But a crawlable site is not yet governed. It may be indexable and ambiguous. It may be visible and misrepresented. It may be fast and poor in action signals.

Level 2: machine-readable site

The site publishes explicit surfaces: structured data, clean routes, sitemap, robots.txt, llms.txt, pillar pages, definitions, manifests, or context files. Systems can more easily find important resources.

At this level, the danger is false completeness. A machine file may orient without governing. Structured data may describe without resolving conflicts. A pillar page may be found without exposing response limits.

Level 3: governed site

The site hierarchizes its sources. It indicates what is canonical, secondary, obsolete, excluded, unspecified, and in need of clarification. Definitions stabilize vocabulary. Proof is connected to claims. Machine surfaces are no longer decorative.

This is where Dual Web becomes explicit: a human surface and a machine surface work together to reduce ambiguity. But the interface may still be weak for action.

Level 4: agent-readable site

The site becomes readable as an action environment. Agents can connect the page, navigation, DOM, accessibility tree, CTAs, forms, errors, and confirmations. Critical journeys are coherent.

This is where visual stability, labels, accessible names, roles, states, and hydration become strategic. A level 4 site lets an agent understand what it sees and where it is.

Level 5: agent-actionable site

The site no longer merely becomes readable. It exposes actions whose intent, parameters, consequences, confirmations, and limits are understandable. Sensitive actions cross an execution boundary. Action legitimacy becomes a condition, not a deduction.

This level does not mean that everything must be automated. On the contrary, it requires legitimate non-actions: asking for precision, refusing inference, escalating to a human, or limiting action when authority is missing.

Quick diagnosis

The right question is not “what score does the site get?”. The right question is: what minimum level is required for each journey?

  • An article page may target level 3.
  • A service page with CTA may target level 4.
  • A diagnostic form may require level 5.
  • A sensitive doctrinal page may require strict level 3 with explicit non-actions.

Usage rule

Use this model with the AI visibility, machine discoverability, and agentic readiness matrix and the Agentic Web Readability Framework. Agentic maturity does not replace SEO, accessibility, or governance. It connects them at the moment when the Web becomes actionable.