Operational Safeguards
What we can prove,
and what we're still building.
We believe transparency is a prerequisite for corporate trust. Below is an honest, verified accounting of our security architecture, data gates, and current certification pipelines.
In Place Today
- Workspace isolation: Enforced at the database layer (not the application code) via triggers and row-level rules, preventing data leakage across workspaces.
- Immutability triggers: Database-level constraints prevent any retro-active alterations or deletes to logged execution records.
- Automated build checks: Build pipeline automatically fails the test suite if any new schema table is registered without a strict tenant ownership constraint.
- Credential safety: Third-party API keys and integrations are referenced as environment mappings, never stored directly in readable database tables.
- Human approval loop: Complete gate blockade requiring an explicit human click-off in the workspace before any message publishes or contacts a customer.
Building
- Per-run spend ceilings: Automated API budget capping per execution thread to eliminate runaway execution costs.
- Automated PII masking: Client-side scrubbing of sensitive customer fields before tokens are dispatched to LLM providers.
- Full audit export: Immutable export format of workspace logs signed on-chain for external audit teams.
Not Yet
- SOC 2 Compliance: Not yet started. We state this plainly rather than suggesting otherwise. Audit pipeline scheduled for Phase Three.
- Self-hosted deployment: The platform runs fully managed; VPC and on-premise deployments are currently scheduled as later-stage roadmap items.
- SSO / SAML Integrations: Workspace access relies on OAuth and magic link authentication. SAML-based identity systems are unscheduled.
* Note: Every competitor leads with compliance badges. We state plainly what is and isn't in place — a guarantee built on triggers and ledger assertions, not self-reported promises.