Documentation

Orvay is the control plane through which humans and AI agents jointly run a company. You give it goals, context, integrations, permissions, budgets and policies. It decides what matters next, delegates it, executes it, verifies it with a different actor, and leaves a record you can check.

The one claim everything follows from

An agent saying "done" is not proof that something is done.

Every design decision in this product is downstream of that sentence. Evidence is immutable and hashed. The actor that proposes an action is never the actor that verifies it. Nothing crosses the boundary to a real system without a contract, and nothing runs at all without passing eight gates in a fixed order.

Where to start

If you are evaluating Orvay, read The eight gates. It is the shortest path to understanding what the product actually does.

If you are integrating it, start with ActionContract, then Capabilities.

If you are deciding whether to trust it, read Evidence and verification. That page is the argument.

What Orvay is not

A chatbot. A coding agent. An analytics tool. A startup generator. A collection of AI personas. If something makes Orvay look like any of those, it is the wrong change.