Security and responsible AI
Built for high-stakes work
Enterprise and public-sector AI depends on security, governance and accountability. We treat them as build requirements. Here is how we work and what you get.
Data handling and privacy
- Deploy in your cloud or region when you require
- Encryption in transit and at rest
- Data minimization: we collect the minimum to do the work
- Your data never trains another organization's model
- Retention controls you set
Access and infrastructure
- Least-privilege access and role separation
- Row-level security on data stores
- OAuth and scoped permissions for connected accounts
- Secrets isolated from client code
- Our own analytics hashes visitor identifiers and stores no raw IP
Evaluation and assurance
- Evaluations before release and on a schedule
- Human review in the loop for high-stakes decisions
- Monitoring, logging and audit trails
- Red-team review of prompts and access paths
Responsible AI
- Answers carry citations you check
- A named person stays accountable for every decision
- We flag low-confidence output for review
- We tell you when AI is the wrong tool
Governance and procurement
- SOC2-aligned controls in our products
- Security documentation for your procurement
- Named, accountable delivery leadership
- Clear data processing terms
What you get
Every engagement includes security and assurance.
- A threat model for your system
- A security review before launch
- Evaluation results you read
- Monitoring and audit trails
- Documentation for your compliance team
- A named owner for delivery
Need our security documentation?
Tell us your requirements. We prepare documentation for your security and procurement review and walk your team through our controls.