Project management your security team will sign off
Full read/write AI agents on every plan, with the governance to match: a complete audit trail, agent attribution, approval workflows, per-agent permissions and org-wide policy. Built for IT, security and operations leaders who need to give their teams a modern, AI-native work platform without losing control.
AI audit trail + export
Every agent action is recorded with the tool, the person it acted for and the client it came from. Export the MCP audit trail to CSV, or stream it live to your SIEM.
Agent approval workflows
Route selected agent writes to a human for approval before they apply. Approvals are opt-in per tool group, and undo is always the safety net if something slips through.
Agent attribution + identities
See exactly what your AI did: "via Claude" and "via Cursor" chips on every change, plus workspace-owned agent service accounts that are assignable like a teammate and never use a paid seat.
Per-agent and per-client controls
Scope each agent or connected client to specific tool groups and permission levels, add per-key rate limits and IP allowlists, and set org-wide policy defaults that new workspaces inherit.
Kill switches + email controls
Disable individual MCP tools, lock whole tool groups org-wide, and gate outbound email actions per workspace with optional domain whitelisting for incident-mode safety.
AI quotas + role-based access
Cap AI spend per workspace and per user, and gate everything behind workspace and project roles with read/write/admin levels and granular guest access.
Common procurement questions
Do you support SAML/SSO?
On the roadmap for Q3 2026. Talk to us if you need a firmer date for procurement.
Where is data hosted?
Azure UK (West Europe / UK South). EU and US-East regions on request.
Can I download a DPA?
Yes. Request via /contact. We send the latest GDPR DPA as a signed PDF.
Do you have audit logs?
Yes. The unified audit log covers automation, wiki, projects and tasks, and a dedicated MCP audit trail records every AI agent action. Both export to CSV, and the AI trail can stream to a SIEM.
How do we control what AI agents can do?
Full read/write MCP is available on every plan, and control is where Business earns its keep: per-agent and per-client tool-group scoping, permission caps, rate limits, IP allowlists, approval workflows for sensitive writes, and org-wide policy defaults. Every write is attributed, previewable and undoable.