Your mail, handled — and still private.
A confidential virtual secretary that runs on a device you own.
What you can do today:
- Create an account and sign in, here on this site.
- Build the Bot from source and run it on your own machine.
- Claim a Bot from your browser, and pair the two directly.
Designed and specified, but not usable yet:
- Mail and messaging
- The end-to-end encrypted AI
The privacy page says exactly which parts are which, and why.
Aida's purpose is to read your mail and messages, work out what matters, and draft the replies you would have written — on a small device you own, in your home, not in somebody else's account.
Read the setup guide or create an account.
Why Aida
Planned It will work while you do not
The plan: mail triaged, WhatsApp summarised, replies drafted and tasks raised continuously, rather than when you remember to open an app. None of it is built yet.
Today Your Bot is yours
This part works now. The Bot runs on hardware you own and holds its own keys, and the server that carries its traffic forwards ciphertext it has no way to read or usefully store.
Today Accounts, and very little in them
You can create an account and sign in. What that account holds is set out in full on the privacy page, table by table; none of it is the content of anything you write.
Planned Encrypted end to end, including the AI
The design: prompts and replies encrypted to a confidential-computing enclave whose attestation your Bot verifies itself, so a model provider that cannot decrypt your message cannot train on it either. Both halves are now written and tested, Aida's and your Bot's — but nothing in the Bot you can run calls them, so no prompt is ever sent and none of this protects you yet.
How the privacy works
Aida is built so that we cannot read your messages even if we wanted to, and so that you do not have to take that on trust. See what each part of the system can see, and which of it is running today.