Trust
This page says what StudyWeave does today to keep your records whole and your decisions yours, and nothing it doesn’t do yet. Each control below maps to a Part 11 clause and a written requirement.
sha256 3f9a…c21eTwo-factor sign-in for everyone
An append-only audit trail, chained so any change shows
Signatures bound to what was signed
We don’t train AI models on your protocols
Records and signatures
The controls behind every study map, form and approval, in plain language.
An append-only audit trail, written in the same step as the change itself. Earlier versions are kept, never overwritten, and each entry is chained to the one before.
Each signature shows the signer’s name, the time and what it means, and is bound to the exact content signed. Change the content and the signature no longer matches.
A password and an authenticator to sign in, and your password again for every signature. Sessions end after 30 idle minutes.
Roles are set per study, and every action is checked twice: by the application and by the database itself.
No forms before the study map is approved; nothing published without the required signatures.
Export records, the audit trail and signatures as JSON, CSV or PDF, readable by people and by machines.
StudyWeave gives you the controls. Validating it for your intended use stays with your organisation, and we share the evidence we have to support it.
Security and data
We don’t train AI models on your protocols.
Database rules keep every organisation’s and study’s records apart, on every table.
Admins can’t set your password or read your authenticator. An account reset by our team is written to the audit trail.
It works from your protocol and your study, and every run is recorded.
Data travels over TLS and is stored encrypted (AES-256) by our hosting providers.
StudyWeave works with protocols and study designs, not patient records.
Quality and change control
Each change is written up before it’s made.
Every change goes through a pull request.
Reviewed before it’s merged.
Automated checks run on every change.
An updated traceability matrix.
How we evaluate Loom
We test Loom on a set of held-out protocols that are never used to tune it. We’ll publish the method alongside the results.
Subprocessors
| Role | Provider | Purpose | Location |
|---|---|---|---|
| Database and file storage | Supabase | Study records, protocols, audit trail | United States |
| Application hosting | Vercel | Running the StudyWeave web app | United States |
| AI model providers | OpenAI or Anthropic, as configured | Loom’s reading and drafting | [founder to confirm] |
Our Part 11 controls matrix, written requirements, risk assessment, validation approach and traceability matrix, available on request. [founder to confirm the list before launch]