Skip to main content

Rebuilding the Web for AI Agents

Web-scale Infrastructure for the AI-first internet

The open web wasn't built for AI agents. Content is bloated, noisy, hard to access, and there's no standard way for agents, publishers and merchants to interact - or transact. We're fixing that.

Our infrastructure products give enterprises, builders, creators and curators the platform to make data, content and knowledge discoverable, accessible, and economically valuable in an AI agent world.

  • Enterprises - slash your time to AI market
  • Agent builders - get instant access to a RAG internet
  • Creators - leverage your visibility, get paid

1. Trusted Gateway

 

Our Trusted Gateway is high-performance content infrastructure that fetches, renders, and converts any web page to clean, structured markdown. AI agents get consistent, safe web access - without touching origin servers directly or wrestling with raw HTML.

For Builders

The context, the whole context and nothing but the context. In a format (markdown and JSON) that your agents can use immediately. From one API or MCP server.

  • Real-time content, not databased page-snippets.
  • Handles javascript, bot directives, bad markup & format negotiation.

For Enterprise

Make your existing content, products, systems and processes available to AI agents. In less than a day.

  • Full visibility into how AI agents are using your sites.
  • Enforce your your terms-of-use and bot directives.
  • Amplify your AI value with third-party content.

2. Edge-Search

 

Designed just for AI Agents - it's RAG for your corner of the internet

Define an "edge" of the internet that matters to you and selected AI agents. Our Edge-Search product performs scoped retrieval across those sites using a combination of vector, full-text and ontological search.

Edge-Search is designed from the ground up for retrieval-augmented generation (RAG) - not adapted from a general-purpose search engine. Results are returned as right-sized context ready for immediate use by your agent's LLM.

  • Server-side context-seeding and configurable re-ranking.
  • Powerful MCP tooling and reporting.

3. Affinity

 

Community For Your People and Their Agents

An Affinity Group is a cluster of people with a common interest, objective or behaviour pattern. A family, place-of-work, school, purchase journey, holiday plan, sport, club, neighborhood, partner search, health journey... up to 95% of task-oriented activity can take place in, and can be supported by, an Affinity Group.

Our Affinity product turns the power of "what we have in common" into an architecture for The Agentic Web. Affinity uses Edge-Search to curate a set of group web-resources and gives you tools to capture conversations that build on those resources.

  • Town-sized or country-sized groups with geographical controls.
  • Safety and privacy by design.

Ready to build?

 

Whether you're building AI agents that need trustworthy retrieval, publishing content that deserves to be found, or running infrastructure that has to scale - we can help.

Book a Meeting

FAQ

 

What's an Edge-Search Manifold?

A manifold is a curated collection of web properties assembled around a theme, interest, or purpose - think of it as a focused, expert-curated slice of the web. Manifolds are the fundamental unit of organisation in Affinity. All Edge-Search queries are scoped to a manifold, so results come only from trusted, pre-indexed content.

How do AI agents authenticate?

Affinity implements the OAuth 2.1 standard with PKCE support. Developers register client applications, receive credentials, and obtain access tokens that grant scoped access to specific manifolds. Scopes cover search, content retrieval, gateway access, indexing operations, and admin functions.

What does the Trusted Gateway actually do?

md-proxy fetches any web URL, renders it if needed (using headless Chromium for JavaScript-heavy pages), converts it to clean structured markdown, and returns it to the caller. It caches aggressively, blocks SSRF attacks, respects robots.txt, and tracks content changes so Edge-Search can re-index automatically. The result: AI agents get consistent, safe web access without any of the messiness of the open web.

How is Edge-Search different from a standard search engine?

Edge-Search is designed specifically for retrieval-augmented generation, not general web discovery. It combines vector (semantic) and full-text (lexical) search using Reciprocal Rank Fusion, returns token-limited results that agents can use directly, and scopes everything to a manifold. There are 10 MCP tools covering search, content fetch, digests, and entity tracking.

Who is this for?

Content creators and publishers who want their knowledge indexed and monetised through AI channels. AI agent developers who need reliable, curated retrieval instead of raw web access. Enterprises that need multi-tenant, auditable infrastructure with team management and access controls.

How do I get access?

The platform is available now. Get in touch and we'll work out what setup makes sense for your use case.