Security at User Evaluation
The standards your
security team cares about.
SOC 2 Type II, ISO 27001, ISO 42001, GDPR, HIPAA. Zero AI training on your data. The confidentiality bar that big research teams have to hit, hit.
Compliance
Five standards.
Five clean reports.
ISO 42001 certified. Our AI governance framework documents how we build and operate AI features.
ISO 27001 certified. The international standard for information security management.
SOC 2 Type II audited annually for how we store and access your data.
GDPR compliant. EU privacy rules apply to every workspace by default.
HIPAA-compliant policies and controls for protected health information.
How we handle your data
Four commitments
that don’t change with the plan.
Zero AI training on your data
Your Customer Data is never used to train any AI model, ours or anyone else's. PII is masked before any LLM request leaves our infrastructure.
Encrypted at rest and in transit
TLS in transit, AES-256 at rest. Database, object storage, and message queues are all encrypted with rotated keys.
SSO, SCIM, and audit logs
SAML SSO and SCIM provisioning on Plus. Every workspace action is recorded with actor, target, and timestamp.
US-hosted, with EU region available
Production runs on US infrastructure by default. EU-resident processing on request for teams with data-residency requirements.
Your data does not train AI models.
Not ours. Not the LLM provider’s. Not anyone’s. We use commercial AI providers with zero-retention agreements and route every request through our own PII-masking layer first.
Before the request leaves
Names, emails, phone numbers, and account identifiers are stripped from prompts. The agent receives placeholders; the response is re-hydrated on our side.
After the response comes back
No request, prompt, or response is logged for model training. The LLM provider deletes inference data on a 30-day rolling window.
Day-to-day practice
What we do, on a normal week.
- Annual third-party penetration tests on the full app and API
- Quarterly internal vulnerability scans, patched on a 7-day SLA
- Background-checked engineers; least-privilege production access
- Incident response runbook with 30-minute paging SLA
- 30-day delivery log on every webhook; signed with HMAC-SHA256
- Data deletion within 30 days of workspace cancellation
US by default. EU on request.
Production runs on US infrastructure by default. EU-resident processing is available for customers with data-residency requirements. Talk to us about your specific contract clauses; we ship most of them as-is.
Talk to security.
Cert PDFs, the latest pen-test summary, and a vendor questionnaire are one email away.