Backlight connects GitHub Copilot and Cursor usage, with Claude Code and OpenAI in beta, to the commits, pull requests, and bugs in your repos. You see AI tool spend, idle seats, and how your team's output compares with its own baseline from before AI.
Free for up to 10 engineers · read-only access to your code · we never store source code or diffs
Each vendor dashboard reports usage for its own tool. It can’t tell you whether that usage turned into merged pull requests or new bugs, or what the same engineer spends across your other AI tools.
Teams add tools faster than they retire them, and at renewal someone still has to answer whether the spend is paying off.
longer to finish tasks for experienced developers using AI, in one controlled study
Nothing to install on engineers’ machines. You connect accounts your organization already has, and Backlight does the matching.
Add your source control (GitHub, GitLab, Bitbucket, or Azure DevOps), your issue tracker (Jira, Linear, or GitHub Issues), and admin access to the AI coding tools you pay for. Access to your code is read-only.
Backlight links each engineer’s AI tool accounts to their commits, pull requests, and issues, across every connected vendor. It identifies AI-assisted work from the shipped changes themselves, with a confidence level for each. We work from metadata and never store your source code.
See AI tool spend, idle seats, and how AI-assisted work compares with your own baseline from before AI, with adoption and seat spend broken down by team.
See seat spend by engineer, team, and seniority. Vendor totals also include available usage charges, with estimates where actual costs aren't available.
See the paid seats nobody has used recently and what each one costs you every month, so you know what to cut before the next renewal.
On the Team plan, set your own thresholds for churn on AI-assisted code, regressions, bug resolution time, and idle-seat spend. Alerts arrive by email or Slack.
Compare how much AI-assisted and other code gets deleted relative to what was added, and, on the Team plan, track regressions per thousand lines merged.
Lead time for changes, change failure rate, and time to resolve bugs, trended over time so you can see whether delivery is getting faster as AI use grows.
The dashboards fill in as your tools sync. Where there isn’t enough data yet to trust a number, the dashboard says so.
Seat spend by engineer, plus totals by AI tool that include available usage charges. Some costs are estimates.
AI-assisted commits and pull requests per engineer, compared with your own baseline from before AI.
Churn on AI-assisted changes next to other changes, plus delivery and stability measures for the whole codebase.
Dollar figures on dashboards leadership can read, with a note on how each headline number is calculated.
Connect your source control, your issue tracker, and the AI coding tools you pay for. Access to your code is read-only, and we never store source code or diffs.