Security at Subro Genie
Subro Genie helps subrogation firms find and pursue recoveries on auto physical damage claims. The claim files carriers send us are confidential, and this page explains how we look after them.
Where we are today. We do not yet have a SOC 2 report, and no independent penetration test has been done yet. Both are on our roadmap below. The controls on this page are in place in our code and cloud setup today unless marked Planned.
Hosting and data location
- Hosted on Amazon Web Services (AWS) in the US East (N. Virginia) region. Customer data and its backups stay in the United States.
- Managed AWS services throughout, with no servers of our own to patch. The cloud setup is written as code, and automated tests check its security settings on every change.
Encryption
- HTTPS only, everywhere. The database and file storage refuse unencrypted connections.
- The database, backups, claim files and stored secrets are encrypted at rest with a customer-managed AWS KMS key that rotates every year.
- Passwords are stored only as salted scrypt hashes, so nobody can read them.
Access control
- Two-factor sign-in for every user: a password plus an authenticator-app code. Accounts lock after repeated wrong tries, and idle sessions end on their own.
- Role-based access. A person approves every demand letter, and only admins manage users.
- Per-carrier walls: each person sees only the carriers they are assigned to, checked on every request.
- Least-privilege cloud roles. The database is never exposed to the internet, and no long-lived cloud access keys are used.
Monitoring, audit and backups
- An audit log records sign-ins, changes, approvals and downloads, with who and when.
- Automated alarms watch for errors, failures and unusual load.
- Automated threat detection watches our cloud account, and every administrative action in it is recorded and kept for a year.
- Continuous database backups with point-in-time restore (35 days), and versioned claim files. We drill restores quarterly; the latest drill (November 2026) passed.
Secure development
- Every change goes through a pull request, with automated tests passing before it is merged.
- Pinned dependencies, weekly vulnerability audits, automated update proposals, and image scanning.
- A strict Content Security Policy and standard security headers on every page.
- Real claim data is never used in development or testing.
AI and your data
- In production, AI runs through Amazon Bedrock inside our own AWS account, so claim data doesn't leave it.
- Your data is not used to train AI models.
- The AI drafts and cites its evidence. A person approves every letter, and the app never sends anything on its own.
Incidents
We keep written procedures for incidents. If we confirm an incident affecting a customer's data, we intend to notify that customer without undue delay, and within 72 hours of confirmation as our target.
Roadmap
| Item | Status |
|---|---|
| Security overview and pre-filled security questionnaire | Done |
| Cyber liability and technology E&O insurance | Planned |
| Independent penetration test, with a shareable summary letter | Planned |
| Written security policies and incident response plan | In draft |
| SOC 2 Type II | Planned |
Carriers can ask for our full security overview and completed security questionnaire through Venture Subrogation.