Security and compliance.
Built for teams that need practical AI automation with clear security boundaries and auditability.
EU data residency by default
Persistent customer data is stored in the EU (Stockholm, Sweden), including memory, files, and service logs.
No model training on customer data
Customer data is not used to train public or shared models. Each employee runs in an isolated environment.
Transparent operation
AI employees identify as AI in communications. Actions are logged and traceable.
Isolation by design
Each employee runs in a sandboxed container with dedicated sessions, memory, and access controls.
Defense in depth.
Multiple security layers keep AI employee execution within clear operational boundaries.
Single-manager trust model
By default, only a verified manager can issue instructions. Unauthorized attempts are blocked and logged.
Sandboxed execution
Employees run in restricted containers with controlled permissions, resources, and file access.
Tool-level access boundaries
Tool access is explicit. Network, shell, and file operations are policy-constrained and logged.
Prompt injection defenses
Layered controls protect instruction channels, including instruction-data separation and structural checks.
Audit logging
Actions are recorded with timestamps and context. Logs are visible through the dashboard.
Data lifecycle controls
Data export is available on request. After termination, data is permanently deleted within 30 days.
Regulatory compliance.
GDPR
Built in- Persistent data hosted on EU infrastructure (Stockholm)
- International transfers covered by SCCs and DPA safeguards
- Standard DPA included for all customers
- Deletion within 30 days after termination
- No customer data used for model training
- Data export available on request
Transparency
Default- AI identity disclosure in external communication
- Audit logs for every action
- Human oversight through a single-manager model
- No customer data used for training
- Observability dashboard per employee
Security questions?
We can walk through controls, architecture, and data handling in detail.