How AIV handles your data and your records
A quality team asks these questions before it trusts any tool with GxP records. Here are our answers, and what is still being built.
Your data trains nothing
Nothing you put into AIV trains a model, ours or anyone else's.
A person makes every decision
The AI layer drafts. It never classifies a system, approves a result or lowers a criterion.
Roles are separated
The person who runs a test can't approve their own result.
Part 11-style e-signatures
Each signature carries an identity check, a name, a role, a meaning and a timestamp.
A full audit trail
Every change is recorded. A record can be voided with a reason, and nothing is ever deleted.
Separate customer records
Each customer's records are kept apart, with a login and a role for each user.
This is what an inspector sees in the audit trail
Every action is written to the trail with who did it and when. The 2 refused actions are recorded too, so an inspector can see the rule held.
Illustration with sample data. The full audit trail arrives with the MVP.
- Acceptance criteria draftedDrafting layer
- Draft corrected and accepted by R. Mehta, QA reviewerReviewed
- Change to lower a criterion by the drafting layerRefused
- OQ result recorded by S. Iyer, testerPass
- Approval of own result by S. IyerRefused
- Release signed by A. Rao, Head of QA: approved for releaseSigned
We validate AIV the way we ask you to validate your systems
AIV's own validation runs from the first day of the MVP build. An independent validator reviews it.