Trust.

What we do with your data — stated plainly. Most companies tell you your data is safe. Fewer tell you what they can actually see. This page does both.

We hold the keys.

We encrypt your data at rest with operator-managed keys. Every time we decrypt your data to work on it while you are away, it is logged in your dashboard. We won’t read your data — and we make that background work visible to you.

In full, this is what you agree to when you sign up:

“In short: Brainontic encrypts all your data on disk. We hold the technical keys to unlock your data — that is needed to do your background work (summarizing, making it searchable) while you are offline. Every time we unlock your data while you are not active, we record it in a log you can inspect. We do not use it for any other purpose, and we make everything we do transparent.”

In plain terms:

• We hold the keys. That is what lets Brainontic work while you are offline.
• We don’t read your data. Not for training, not for analytics, not out of curiosity.
• You don’t have to take our word for it. Every time our servers unlock your data to work on it without you, a row appears in your own activity log.

Why it works this way.

Brainontic does its best work while you are not there. Overnight it connects what you added to what you already had, clusters related material, writes summaries, and keeps your index current. That work has to decrypt your data at a moment when you are not present to supply a key.

We could have built a zero-knowledge system — encrypted in your browser, unreadable to us. It would also have been unreadable to your agents, unsearchable in the background, and unable to think while you sleep. We chose a brain that works over a vault that doesn’t. The longer argument is in our manifesto.

How you check us — trust, but verify.

Open Dashboard → Security. Every time a background job decrypts your data, it writes a row there:

• what happened — entity-stitching, community update, RAPTOR summary, connector sync, email ingest
• when, in UTC
• how many items were touched
• which worker or token did it
• which key scope was used

The page shows the last 30 days. It records that an operation happened and what kind — never the content itself.

The same rows are included in your data export, so you can keep your own copy. If you ever see background activity that does not match what you did, you have evidence immediately — not after an investigation.

What we never do.

• We don’t train models on your data — ours or anyone else’s.
• We don’t sell it, rent it, or use it for advertising.
• We don’t mix it with another customer’s. Every tenant has its own encryption key; one customer’s key cannot open another customer’s data.
• Our connectors only read. Brainontic never sends, changes or deletes anything in your Gmail, Drive, Dropbox, OneDrive or Notion.

Where your data lives.

On dedicated hardware in Falkenstein, Germany — inside the EU, under the GDPR, on a machine that runs nothing but Brainontic. Not a shared cloud tenancy alongside strangers.

Who else touches your data.

Brainontic cannot work alone. These companies process part of your data on our behalf:

• Anthropic — receives your questions, the passages retrieved to answer them, and the text of what you store. Writes your answers, and the summaries and labels that make your brain searchable.
• Google — receives the text you store and every question you ask. Turns both into the numbers that make search work.
• Stripe — receives your name, email and payment details. Takes the payment; never sees your knowledge.
• Nango — the software that brokers your connector sign-ins. We run it on our own hardware, so your access tokens stay with us and never reach Nango the company.

Behind them sit the providers any online service needs: the company whose hardware your brain runs on, and the relay that delivers your sign-in links. We do not add a processor without updating this page, and if you need the complete current list in writing — for a data-processing agreement, or because your own compliance team asks — request it at brain@brainontic.ai and you will get it.

Deleting everything.

Dashboard → Delete account. Your documents, your knowledge graph, your memories, your connector credentials, your agent tokens and your sign-in are erased, and your encryption key is destroyed — which leaves any remaining ciphertext unreadable by anyone, including us.

Some records survive, and we would rather describe them than count them. Your account row is kept as a closed-account record for tax law: the closure date, your plan and subscription state, when you signed up and last signed in, and a billing reference to your payment history at Stripe. Your name, email and sign-in identifiers are erased from it. It carries no knowledge of yours and no credentials — but it is not anonymous, because that billing reference could still connect it back to you. We would rather say that than call it anonymised.

Two operational logs are kept as well. The background-activity log stays deliberately — erasing our own accountability trail the moment an account closes would be the wrong way round; it records what ran, when, how many items and which worker, and no knowledge of yours. A separate internal access log is kept for up to six months, and it is the one exception worth spelling out: alongside the request it records the first 256 characters of questions and searches, so for that window a fragment of what you asked can outlive the account it was asked from. That window is a setting rather than a law of physics; if it matters to your organisation, say so and we will talk about it.

You can take your data with you first — Dashboard → Export data packages up your documents, your graph, your memories and your activity log. It has limits, and they are real: very long documents are cut at 50,000 characters, and conversations, contacts, communities and usage history are capped per section, so a large brain will not come out whole. Above a hard ceiling the export refuses outright rather than handing you a partial copy that looks complete. If you need everything, write to brain@brainontic.ai and we will get it to you.

Where the boundaries are.

Every architecture makes a choice, and every choice has edges. Ours are worth knowing before you commit.

• We hold the keys, by design. It is what lets your brain keep working while you sleep. It also means our access is governed by policy and evidence rather than by mathematics — which is precisely why the activity log exists, and why it is yours to read.
• The log lives where your data lives. It is a complete record of the background work we did on your data, kept on the infrastructure that did it. It does not log the decryption that happens while you are using Brainontic yourself — that one you are present for. If your organisation needs audit anchored outside its provider, write to brain@brainontic.ai — that is a different setup, and we will tell you plainly whether it fits you.
• Encryption at rest is protection against a stolen backup. Our off-site copies are encrypted with a key that is not kept on the server, so a stolen backup is unreadable. A stolen production server is a different matter: the master key lives on it, because the service needs it to run. That is the honest limit of encryption at rest — here, and nearly everywhere.

Questions.

If anything here is unclear, or your organisation needs a data-processing agreement, write to brain@brainontic.ai. We would rather answer a hard question than win you with a vague page. The terms themselves are on this page, readable in full whether or not you have an account.