It builds real UI
The agent doesn’t describe your product — it operates it through your own API, as the signed-in user, and renders the screens, views, and micro-apps it generates in your brand, in a sandboxed surface.
It's the full stack
Knowledge bases, user context — it knows who’s signed in and sees what they see — the harness, API tools, guard and approvals, automations. Included, composable, yours to swap.
Pick what you already have
You already have an agent
Recommended. AI SDK, Mastra, or a loop you wrote yourself. One tool
pack spreads into it and your agent starts acting through your API and
building real screens.
Your product has no agent
Beta. Vendo brings the whole thing — the loop, the chat UI, the
approvals. One command wires it into your app and your users are talking
to your product.
Expose your product over MCP
Experimental. Claude, ChatGPT, Cursor, and Claude Code connect over
OAuth and act as the person who signed in — the same guard, approvals, and
audit as your own surfaces.