The Audit Trail: what the AI shows you every time it touches your notes

Ask most AI assistants to do something and you get a result. You rarely get a real accounting of what it actually did to arrive there — which of your data it looked at, what it considered doing and didn't, or why it made the specific choice it made. That is a reasonable trade for a search engine. It is a much harder trade to accept for something with standing access to your private notes.

Receipts, not just results

Every action Prevon's on-device AI takes — what it did, when, and why — is logged somewhere you can actually open and read, in plain language rather than a raw technical timestamp. Which notes were touched, what the AI decided to do, and the reasoning behind it. Not a debug log meant for engineers. A record meant to be read by the person whose notes it touched.

Most AI agents give you a black box. This gives you receipts.

Why this only works because the AI is on-device

A cloud-based AI feature could, in principle, also log its actions — but the log itself would live on someone else's server, alongside a record of what your AI assistant was doing inside your private data. Because Prevon's AI runs entirely on your device with no code path to the internet, the audit log lives in exactly the same place your notes do: nowhere but your device, under nobody's control but yours.

Paired with the permission tiers covered elsewhere on this blog — read access is automatic, anything that changes or deletes something needs your confirmation first — the audit trail is what turns "trust the AI" into "verify what the AI actually did." You should not have to take either claim on faith.