We're helping write the standard for AI on the web
The Web Model Context Protocol (WebMCP) is an emerging W3C standard that lets AI agents act on a web page, not just read it. No separate or extra infrastructure required.
The problem today
AI agents interact with your site by blindly clicking and scraping. It breaks constantly. Connecting agents the "right" way means building and maintaining a whole separate backend. That creates three compounding risks:
- Scattered credentials. Every integration needs its own authentication. API keys, service accounts, and OAuth tokens end up spread across systems. WebMCP works with the browser session the user already has.
- No accountability. Audit logs say "Agent," not who triggered it. WebMCP ties every action to a real identity.
- Months of integration work. You're rebuilding logic your frontend already has. WebMCP wraps what's already there.
Auditable by design
The browser controls what agents can do. Users see every action, approve or deny it, and stay in control. Every tool call is traced, replayable, and auditable.
Where SigVelo fits
SigVelo helps write the WebMCP specification at the W3C, alongside Google and Microsoft. Chrome and Edge already ship an early implementation. We're building the tools to get your site ready before the standard lands in every browser.