Total RecallPrivacy
Your memory is yours.
Total Recall is built around one architectural promise: your brain, meaning the database that holds your documents, messages, contacts, and everything derived from them, runs in accounts you own. We designed the product so we cannot read it.
Effective date: July 12, 2026
Where your data lives
When you set up a brain, the setup wizard deploys it into your own accounts: a Postgres database (the storage engine) in your Supabase account, always-on services in your Railway account, and AI processing billed to your own OpenAI and Anthropic keys. Your documents, emails, chat messages, contacts, and the memory built from them are stored there, under your logins, on infrastructure you can inspect, export, or delete without our involvement.
What we store
The theo app (this website and the setup wizard at asktheo.totalrecallhq.com) keeps the minimum it needs to function: your account identity from sign-in (name and email, via our sign-in provider Clerk), your setup progress (which steps you completed and your setup questionnaire answers), and standard operational logs. That is the list.
What we deliberately do not store
The wizard is a zero-credential router. The API keys, database passwords, and tokens you paste during setup are used transiently, meaning we check them and pass them directly to your own infrastructure without writing them to our systems; between steps they live only in your browser tab. When you connect a source through a Total Recall hosted app (for example our Slack app), the resulting access token is delivered straight to your own services and is never stored on ours. Your brain's contents never pass through our servers at all: search and chat happen between your devices, your brain, and your AI provider accounts.
The services involved
Running theo involves a small set of third parties, each with a narrow job: Clerk (sign-in), Vercel (hosting this app), and the providers of accounts you own and connect yourself, such as Supabase, Railway, OpenAI, Anthropic, Google, Slack, Notion, and Telegram. Sources you connect are governed by the permissions you grant on those platforms, and you can revoke them there at any time.
AI processing and training
Your content is processed by AI models using your own OpenAI and Anthropic accounts, under those providers' API terms, which by default do not use API content to train models. We never use your content to train anything, and we could not: we do not have it.
Deleting your data
Because your brain lives in your accounts, deletion is genuinely in your hands: deleting your Supabase project and Railway services removes the data and stops the billing, and our guides walk through an orderly export first if you want one. To remove what we hold, email us and we will delete your account identity and setup records.
Changes and contact
If this policy changes, we will update this page and its effective date. Questions, deletion requests, or anything unclear: support@totalrecallhq.com.