Heads of AI

One record of every agent’s work, across every team and every vendor.

“We run many agents across teams and vendors. Who governs them?”

Your teams already use several AI vendors, and more keep arriving. FEATit governs all of them the same way, in one place, without tying you to any one of them.

AI agent governance, in full

How it works for you

  1. Any vendor, one record

    Leading-edge cloud agents, cost-efficient models, local models and your own AI platform connect through the standard agent connector (MCP), a command or an API. Their work lands in one history, named the same way whoever made it.

    Which AI setups connect, and how
  2. Who may do what, from the organisation down

    Each person and each agent reaches only what its roles allow. Roles are granted for the whole organisation, a branch of it or one team, and an agent signs in with its own credential, scoped to its roles.

    How permissions are set
  3. The organisation mirrored, each team its own process

    Business units, products and teams each keep their own views, statuses and rules, and access passes down unless a team opts out. One governance model, many ways of working.

    How teams shape their own process
  4. Your record, in our cloud or on your servers

    Run FEATit in our cloud or on your own infrastructure. Everything in a work area exports as CSV and JSON whenever you want, so the record never depends on one vendor.

    How your record stays yours

Every vendor, connected the same way

Ready-made adapters reach the cloud agents your teams already use, each with one thing to connect. Anything that can use the standard agent connector (MCP), a command or an API connects too.

Example flow
Which adapter reaches which agent
AgentAdapter
Claude Code RoutinesWebhook
Claude Code GitHub ActionGitHub issue
Codex cloudGitHub issue
Cursor Cloud AgentsWebhook or Vendor API
GitHub Copilot cloud agentGitHub issue
DevinVendor API
OpenHandsGitHub issue
JulesVendor API

A week with FEATit, as head of AI, across two teams

Example flow
  1. Monday

    Team Payments runs Claude Code on its Macs, Team Mobile runs Codex. Both work under one organisation, with the same roles and one record.

  2. Tuesday

    Team Mobile adds a cloud agent through the GitHub issue adapter. It inherits the organisation’s roles and the team’s rules on its first day.

  3. Wednesday

    Team Payments tightens its definition of done for payment code. Team Mobile keeps its own, and neither has to ask the other.

  4. Thursday

    Legal asks who approved a change a Team Payments agent made last month. The item’s history names the agent session, the person who vouched for it and the reason.

  5. Friday

    For the quarterly review you export both teams’ work areas as JSON, every item’s history included, whichever vendor did the work.

Questions

Can FEATit govern agents from different AI vendors?
Yes. FEATit is independent of your AI vendors. Anything that can use the standard agent connector (MCP), run a command or call an API takes work and reports back under the same rules.
Can we run FEATit on our own servers?
Yes. FEATit runs in our cloud or on your own infrastructure, with the same product in both.
Do agents get their own identity?
Yes. An agent signs in with its own credential, limited to the roles it holds, and every change it makes is attributed to it and to its session.

Not there yet

  • Coming soon: single sign-on for your verified domain, and SCIM provisioning.
  • Coming soon: administration across several organisations and business units.
  • Coming soon: EU data residency in our cloud.

Other readers