Business plan
What do your AI coding agents change?
Sprint Radar compares AI-assisted work with other work, from your Jira data. It reports what it measured and how many items it measured. It does not claim that AI caused a difference.
How it works
You define AI-assisted work
A project owner or admin sets rules that recognise AI-assisted issues: by label, by assignee (the current one, or any assignee in the status history), by a custom field, or by creator. Without rules, Sprint Radar suggests labels and accounts that look like agents, for you to confirm.
It runs with each forecast
The analysis uses the same Jira history as the forecast. Workspaces without the Business plan get no AI Impact result for their own projects, and nothing is computed for them. The demo project in every new workspace shows an example on any plan.
What it measures
- Items and points delivered per sprint, for AI-assisted work and for other work
- The share of delivered work that is AI-assisted
- Lead time and cycle time, from real status changes
- Rework: how often finished items are reopened
- Abandoned work: items closed without delivery
- Work in progress for each group
- A comparison of cycle time and reopen rate, with a 90% interval
- Throughput in the six sprints before and after an adoption date you set
- A forecast that leaves out AI-assisted completions, labelled as an upper bound
What it refuses to do
- It withholds a comparison when either group has fewer than 20 items.
- It compares items of the same type only when both groups have at least 20 of that type.
- Every comparison carries a caution. The numbers describe what happened. They do not show why.
- It says how many issues have no assignee history, because an “any assignee” rule then sees only the current assignee.
Where it stands
In the app
Open a project and choose the AI Impact tab. It shows the comparison, the sample sizes and the rules editor for owners and admins. The demo project in your workspace shows a populated example.
Preview
Usage and cost for Claude Code, GitHub Copilot and Cursor, on the AI usage page. Connect the Claude Code Analytics API (Admin API key) or send OpenTelemetry metrics, connect GitHub Copilot usage metrics for an organization or an enterprise, or connect the Cursor Admin API. It shows spend, cost per delivered item and rejected edits next to what your projects deliver, with a breakdown per tool. No prompts and no code are read.
What each tool reports
Claude Code and Cursor report spend. GitHub Copilot reports usage, not cost: with a price per seat the page shows an estimate, labelled as an estimate and kept out of cost per delivered item. The Cursor Admin API is listed by Cursor for Enterprise teams. The three connectors are previews: they have not run against many real organizations yet.