AI Governance

Know exactly where your AI budget is going. Every token, every seat, every API call — tied to a person, a team, a project, a cost centre, an outcome.

Most tools tell you what happened last quarter. Flowstate tells you what is happening, why, and gives you the controls to do something about it. Stop flying blind. Then stop the leaks.

This is the AI Governance pillar — one of three in Flowstate (alongside Workforce Engineering and Finance & Compliance). It runs on telemetry from the Flowstate Agent on developer machines, plus authoritative billing pulls from your providers.

Flowstate covers coding tools first-class today and is expanding across image, video, audio, sales, marketing and legal — see the Catalog for what's visible right now.

What's in here

Insights — what your AI estate is actually doing.

• Attribution — every prompt, seat and API call resolved to a person, team and project.
• Spend reconciliation — provider invoices matched against observed usage, Xero-style.
• Forecast — six-month spend projection by project and team, with confidence.
• Fluency — how well your engineers describe, delegate and discern with AI.

Governance — the controls.

• Alerts and signals — behavioural-signal stream covering content safety, brand and usage, budget, and operational categories.
• Providers inventory — every AI tool registered to your org, who's using it, and when it last sent traffic.
• Policy and enforcement — active policy, drafts, version history and audit log.
• Service accounts — production AI workloads governed separately from people.

Agent — the on-device sensor.

• Agent overview — install paths, two pathways (Telemetry vs Enterprise), how the agent ships.

Cloud Proxy — network proxy routing and subject metering for managed fleets.

• Cloud Proxy overview — ai.flowstate.inc, org key provisioning, MDM payloads, and the subject meter.
• Install the cloud proxy — deploy the shipped MDM proxy payload and verify traffic.
• Troubleshooting — pending status, proxy auth, and fleet rollout checks.
• Security and privacy — what the shipped Cloud Proxy stores and what it does not.

Self-service — what your engineers see in their own session.

• My AI Usage — personal standing, nudges, sessions tab.
• My AI Keys — request and rotate the keys for the tools you actually use.

Admin and connections — the pipes.

• Usage providers — connect provider billing APIs (Copilot, Anthropic, OpenAI, Cursor) for authoritative spend.
• Flowstate in Claude — install Flowstate as an MCP server for Claude.ai.
• Flowstate in ChatGPT — same MCP server, ChatGPT Connectors install path.

How it fits together

Read Concepts for the telemetry → signals → policy loop in one page. Then pick the deployment pathway that matches your AI estate at Telemetry vs Enterprise.

We're an insight and policy layer. Not a reseller. Not a billing middleman. We don't touch the money — we just make sure you can see where it's going and decide what to do about it.