REFERENCE · THE AI-GENERATED WEB

The Web generatedby AI

Explore the Web generated by and for artificial intelligence: foundational concepts, technologies, standards, tools, organizations, applications and research. An independent source-led reference atlas.

Portal preview · unreviewed records excluded
REFERENCE PORTAL · WEBAI

The key domains of the AI Web

A structured directory of concepts, standards, technologies, publications, organizations and applications, linked to original evidence.

Encyclopedia
FOUR DIMENSIONS OF WEBAI

One Web. Multiple transformations to document

AI affects creation, information access, agent actions and trust. These are domains to compare, not mandatory stages of one inevitable evolution.

03The agent-facing Web

Interact and act

Agents can retrieve data and call tools through interfaces and protocols. Permissions, human approval and outcome checks remain essential.

STANDARDS AND PROTOCOLS

Interoperability starts with precise specifications

From HTML to agent tools: compare specification scope and status, with direct links to original documents.

Explore the registry ↗
HTML Living StandardWHATWG

Structure and semantics of Web interfaces used by people and agents.

Living StandardSource ↗
JSON-LD 1.1W3C

Expressing linked data and structured relationships inside JSON documents.

W3C RecommendationSource ↗
Model Context ProtocolMCP maintainers

Structured tools, resources and exchanges between AI clients and connected services.

Versioned open specificationSource ↗
Agent2Agent (A2A)A2A maintainers

Task exchanges, messages and descriptive agent cards between agents.

Open protocol; verify versionSource ↗
WebMCPW3C Web Machine Learning Community Group

Proposal for Web page JavaScript functions exposed as tools to agents.

Community Group draft; not a W3C StandardSource ↗
C2PACoalition for Content Provenance and Authenticity

Manifests, assertions and signing mechanisms for media provenance claims.

Versioned technical specificationSource ↗
DOCUMENTED ECOSYSTEM

Standards, projects and accountable organizations

The directory separates standards bodies, open-source projects, security initiatives and content providers. No affiliation is implied.

All organizations ↗
NEW REFERENCE DIRECTORY

AI agents: from assistants to action-oriented systems

AI agents increasingly shape how the Web is researched, navigated and operated. WebAI distinguishes usable agent products from SDKs and frameworks for developing new agents. Every entry links to usage guidance and original sources.

The directory does not rank products best to worst: suitability depends on tasks, permissions and verifiable outcomes.

Explore 22 AI agent profiles ↗
12Ready-to-use agent products
10Agent frameworks and builder platforms
44Individual French and English profile pages
THE WEBAI ATLAS

A reference atlas from sources to technologies

An encyclopedia, tool directory, official documents and an interconnected glossary: every entry links back to its origins.

Search WebAI ↗
NAVIGATE THE REFERENCEConnected paths for different research questions
DECISION SUPPORTDocuments, comparisons and verification
TECHNICAL EXPLORATION

A laboratory for inspecting mechanisms

Fifteen browser-only modules for JSON analysis, MCP tool schemas, HTML and sitemap checks, data comparison and evidence review. They run locally and explain their limitations.

Open WebAI Studio ↗
WEBAI · DOSSIERS DE RÉFÉRENCE

A source-led encyclopedia of the changing Web

The encyclopedia’s 36 dossiers connect architectures, applications, research and primary sources. This selection covers agents, search, local inference, standards and trust. Editorial review remains pending.

Explore topic relationships ↗
Retrieval architectureRAG: retrieval-augmented generation

Retrieve documents to inform a generated answer.

Read the dossier ↗
Agentic workflowsThe Agentic Web

Agents, tools, permissions and control of actions.

Read the dossier ↗
Published protocolUnderstanding MCP

Connect AI applications to external tools and resources.

Read the dossier ↗
Community Group draft — not a W3C standardWebMCP: an interface for Web agents

Expose structured capabilities from a Web application.

Read the dossier ↗
Technical provenanceC2PA and content provenance

Describe the origin and some edits of a media asset.

Read the dossier ↗
W3C Candidate Recommendation Draft (Sep 15, 2026)WebGPU and in-browser AI

Use the GPU to accelerate computation and interfaces.

Read the dossier ↗
Visibility in generative searchVisibility in generative search

Discover and evaluate content inclusion in AI answer interfaces.

Read the dossier ↗
Build websites for humans and agentsBuild websites for humans and agents

Make interfaces easier to interpret without degrading human experience.

Read the dossier ↗
Control search and AI crawlersControl search and AI crawlers

Understand the distinct roles of indexing, search and training crawlers.

Read the dossier ↗
Built-in browser AI: capabilities and limitsBuilt-in browser AI: capabilities and limits

Assess translation, summarization and generation APIs offered by some browsers.

Read the dossier ↗
C2PA provenance and independent verificationC2PA provenance and independent verification

Avoid conflating signatures, attribution and factual truth.

Read the dossier ↗
EU transparency for AI-generated contentEU transparency for AI-generated content

Read transparency requirements without generalizing beyond applicable law.

Read the dossier ↗
EDITORIAL FIRST

A reference grounded in sources

WebAI distinguishes published standards, experimental proposals, research findings and editorial interpretation. Records link to original documents and identify pending review.

Explore our method

Primary reference points

Start with the institutions and projects that define the technologies.