Overview
Relay Docs
Relay is an AI chief of staff for your work tools. It connects to the accounts you already use, reads what is happening in them, and gives you one place that answers the question “what actually needs me today?”
These docs are the written record behind that: what the product does, how it is built, exactly what permission it asks each integration for, and the agreement you accept when you sign in. They are meant to be handed to a teammate, a beta user, or a security reviewer without a translation layer.
Start here
There are two documents, and which one you want depends on what you are trying to find out.
- Handbook — the product and security reference. What Relay is, how it is built, how signing in works, where your data lives, how to connect each app, the exact permissions requested per provider, what the assistant can and cannot do, how to test it, its known limits, and the security model.
- Terms of Use — the agreement behind your account. What the service does, what it deliberately does not promise, what each side is responsible for, and how either party can end it.
Relay only ever reads an account after you authorise it, and every write the assistant prepares waits for you to confirm it. The permissions table lists what is requested, provider by provider, with the reason for each scope.
What’s in these docs
Product & security
The Handbook is one long reference, in this order:
- What Relay is — the product, in plain terms.
- How it’s built — the pieces, and which of them ever holds a secret.
- Signing in and where your data lives.
- Connecting an app and setup, app by app.
- Permissions asked — every scope, per provider, with its reason.
- What the assistant can and cannot do.
- How to test it, known limits, and the security model.
Legal
The Terms of Use are numbered clause by clause so a support reply or a review can cite one precisely, and every clause that carries a real consequence is followed by a short plain-English note saying what it means.
- Beta status — what “beta” actually commits us to.
- AI features & disclaimer — what AI output is and is not.
- Data & privacy, and the Privacy Policy it incorporates.
- Limitation of liability and suspension & termination.
How Relay works, briefly
You connect an account. Relay stores the resulting token encrypted on the server and never in the browser. When a page or the assistant needs data, a server function fetches it live from the provider, summarises it, and discards the raw content — Relay is not a mirror of your inbox.
Anything that would change something in a connected tool is prepared, previewed, and only executed after you confirm it. The full version of that story, including what happens on disconnect, is in the security model.
Relay reads what you let it read, writes nothing without a confirmation, and keeps the summary rather than the source material.
Getting help
Relay is in private beta and is run by one person. If something is wrong, unclear, or missing from these docs, mail relayapphq@gmail.com — that address reaches the operator directly, not a queue.