How it works
Audit. Build. Scale.
A clear, three-stage path from a website that AI agents have to guess at, to one they can use with confidence. You approve every step, and nothing goes live without your sign-off.
-
Audit
It starts with a free snapshot of how AI agents see your website today. The full audit, scoped with you, then maps your customer journeys, ranks them by value and scores each one.
- A map of your enquiry, booking, quote and support journeys
- An agent-readiness score for each journey
- A ranked shortlist of the first WebMCP tools worth building
-
Build
We design clear tool names and input schemas, build them into your site’s own code and test them against real agents.
- Declarative tools on your existing forms, where that is enough
- Imperative JavaScript tools for richer tasks, such as stock checks or multi-branch booking
- Testing in Chrome, Edge and ChatGPT Desktop, where WebMCP is supported in trial or preview today
- A permission review, so anything that changes data needs a person to confirm
-
Scale
We watch how agents use your site, add tools per branch, market and language, and keep pace as the standard changes.
- Logging of agent-triggered actions, which your site can recognise
- New tools as your business grows
- Updates as the WebMCP specification evolves
What we need from you
- Access to your website, or an introduction to your web team or agency
- One working session to agree which customer journeys matter most
- Your sign-off on what AI agents may and may not do
You keep ownership of your website and your code. WebMCP tools run in your site, using your customers’ existing logins. No passwords are handed to any AI.
Start with a free snapshot
A short, plain report on how AI agents see your website today and the top three things to fix. No obligation.