AI is no longer just a sidekick to your developers — it now opens PRs, moves Jira tickets, and burns through tokens on its own.
AI Governance V1
brings everything you need to govern that reality in one dashboard: who uses what, what it costs, what it ships, and now — what your autonomous agents
are doing alongside your humans.The dashboard has grown into four tabs that answer four very different questions, plus a new pipeline for ingesting agentic activity at scale.
>
Where to find it:
sidebar → AI Governance
, or open /ai-governance
in your workspace. Filters and date range work across every tab.
What's new
🤖 New tab — AI Agents
We've added a dedicated
AI Agents
tab to track autonomous agents (Claude, GitHub bots, custom Jira agents, …) as first-class actors on your team. Identify each agent by the user account it logs in with, then measure its output, quality and cost.KPIs at a glance:
- Active Agents— how many agent accounts (the "Bots" team) were active in the period.
- Tasks Completed— tasks owned by an agent that transitioned to a Done-equivalent status (Done / Closed / Resolved / Completed / Released).
- Rework Rate— percentage of agent task transitions that movedbackwardsin the workflow (e.g. Review → In Progress). Lower is better — it signals an agent whose output gets accepted on the first pass.
- Agent vs Human Share— share of completed tasks owned by agents vs humans in the period.
- Tokens Consumed— total Claude Code tokens (input + output) consumed by agent accounts.
On the same tab:
- Tasks Completed Over Time— daily delivery cadence per agent, so you can spot ramp-ups, stalls, and weekend runs.
- Top Agents— leaderboard of agents by tasks completed, with Tokens / Task to expose efficiency outliers.
- Agent Run Log— every task an agent handled in the period, with platform (Jira, GitHub, …), task title, current status, and last update. Useful for spot-checking what agents actually shipped, and for audit trails.
🛠️ Better Claude Code ingestion under the hood
We've reworked our Claude Code data pipeline so per-session activity, model selection, and token usage land faster and more reliably in AI Governance. If you've integrated Claude Code via OTLP, you should see fresher numbers across
Most Used Model
, Daily Usage
, and Tokens Consumed
without doing anything on your side.A reminder of what AI Governance already does
If you haven't opened the dashboard in a while, here's what each of the four tabs answers:
- AI Usage— who uses AI tools, how often, and with what models.
- AI Impact— whether AI-assisted PRs are actually faster than non-AI PRs across coding, review, and deploy stages, plus lines of code, batch size, and AI Intensity.
- AI ROI— whether you're getting your money's worth: cost per active user, seat utilization, idle seats, weekly active users.
- AI Agents(new)— what your autonomous agents are shipping, and at what cost and quality.
All four tabs share filters (date range, projects, work items, teams) and a unified data model — so an agent account counts toward Tokens Consumed on AI Usage, Cost on AI ROI, and Tasks Completed on AI Agents, all at once.
How teams are using AI Governance
- Engineering enablement leads— find your power users and idle seats, redirect licenses, and prove the program is actually moving lead time.
- Engineering managers— see whether your team's AI adoption translates to fewer review cycles or just more code.
- Heads of engineering / leadership— quantify cost per AI-assisted PR, and now cost per agent task too.
- Anyone running an internal agent— verify the agent is actually delivering, and audit the work it touched.
Getting started
- Make sure your AI tools are connected (Settings → Integrations). GitHub Copilot, Cursor and Claude Code (OTLP) are supported today, with more integrations on the way.
- Tag your agent accounts as members of a Botsteam so the AI Agents tab can recognize them.
- Open AI Governanceand pick a date range.
Learn more
- AI Governance dashboard
- AI Governance metrics overview
- GitHub Copilot integration
- Claude Code integration (OTLP)
- AI-assisted PRs
- AI-assisted Commits
- AI Lead Time Comparison
- AI Tool Cost
- AI User Engagement Segments
- AI Intensity
Have feedback or a use case we should support? Reply on Canny, or ping us in your Intercom Messenger inside the app.