01
The audit trail cannot drift
Every change writes its audit row inside the same database transaction. If the change rolls back, the record rolls back with it. The API refuses to commit a mutation that skipped its audit write.
The audit row is written in the same transaction as the change.
Most products log an audit entry after the fact, so the log and the database can disagree. In lube they commit together or not at all. That is a runtime rule, not a policy.
What you get on day one.
02
Permissions down to the row
Organization roles, team roles, and per-user grants decide who reads and who writes. API keys carry their own scope, so a key can never do more than the person who made it.
03
AI is off until you turn it on
There is an organization kill switch and a per-user opt-out. Turn AI off and the product still works, because nothing important is gated behind a model call.
04
You choose what comes in
Connect only the repositories, services, and channels you want in the workspace. AI usage data is redacted before it lands, and people control their own visibility.
We do not have SOC 2 yet.
We are early and we hold no compliance certifications today. If that is a hard requirement for you, we are not the right fit yet. If you want to review the controls above against your own bar, start a workspace and look at them directly.
Start free