AI Impact

Is AI paying off?
See it per merge request.

AI Impact sets every AI-assisted merge request against the rest of your work. Adoption per tool, cost in euros per merge request, how fast the work gets reviewed and merged, and whether repositories that lean on AI fail more often.

Part of Scale-Up, from €18 per collaborator a month · every feature free for 60 days · no credit card

Or start your 60-day free trial →

AI Impact

Agile Analytics AI Impact page, desktop view, light theme

Adoption per tool, cost and return in euros per AI-assisted merge request, delivery speed and change failure rate against the rest. Try it in the free tour →

What it measures

Four questions, answered from your merge requests.

How much of our work is AI-assisted?

The share of merge requests an assistant helped write, per tool and per week. Coding agents such as Claude Code, the Copilot coding agent and Cursor add a Co-authored-by line to the commits they help write; that line is what counts. The page shows how many merge requests it could read, so you know how far to trust the share.

Does AI-assisted work move faster?

Time to merge and time to first approval, AI-assisted against the rest. Compared inside each repository and then rolled up, so a fast repository cannot flatter a slow one. Merge to deploy is shown as context, because it mostly follows your release rhythm.

Does more break in production?

Change failure rate, how often a change fails once it is deployed, in repositories where at least half the merge requests are AI-assisted. Set against the repositories that use AI less.

Is delivery improving as AI use grows?

DORA over time: deployment frequency, lead time for changes, change failure rate and time to restore, week by week, next to the share of AI-assisted work.

These are comparisons between groups of merge requests, not a controlled experiment. A difference tells you where to look, not what caused it. Groups of fewer than five merge requests are left out.

See it in the app

AI Impact in under a minute.

On Pied Piper, a company we made up: the share of AI-assisted merge requests per tool, what each one costs in licences and the hours it saved. The whole product tour shows what sits around it: the health score, error budgets, DORA and sprint flow.

Now click through it yourself →Plays from YouTube, loaded only when you press play. Watch on YouTube

Cost and return

Your AI bill, per merge request.

Enter what your licences cost, or let Enterprise read your vendors’ admin accounts every night. Agile Analytics sets the spend against the AI-assisted work that shipped and the hours your people booked. In euros, with the source next to every number.

US enterprises spent $4 billion on AI coding tools in 2025, seven times the year before. Few of them set it against what shipped. Menlo Ventures, 2025

per tool

Cost per AI merge request

Spend divided by the merge requests an assistant helped write. So Claude Code and Copilot can be compared on what they cost per change.

from your hours

Return from booked hours

Hours per merge request in each person’s AI-heavy months against their quieter months, from the hours they actually booked, valued at your hourly rate. Shown only when enough people take part to make it reliable.

Enterprise

Seats against use

GitHub Copilot and Cursor report their seats and who used them. Twelve seats and eight people active this month is a number you can take to the renewal.

Enterprise

Seat fees apart from tokens

Licences on one side, token and agent spend on the other, per tool and per model. Converted at the ECB rate of the day.

Spend sync reads

  • Anthropic: Claude API and Claude Code
  • OpenAI
  • Google Gemini, through your billing export
  • GitHub Copilot
  • Cursor
  • Any other tool, as a licence you enter

Vendors report use per person. We keep that only to link accounts to teams and never show it: an assistant’s use appears when five or more people used it, a team when five of its members did.

Why measure it

Faster typing is not faster delivery.

−7.2%

delivery stability per 25% more AI use

Documentation, code quality and review speed improved as AI use grew. Delivery stability fell. An association across 39,000 professionals, not a before-and-after.

Google DORA, Accelerate State of DevOps 2024

10–15%

typical productivity gain from AI assistants

The time saved often goes to work of no higher value, so it does not show up as a return. Writing and testing code is only 25% to 35% of the time from idea to launch.

Bain & Company, Technology Report 2025

Setup

Four steps. Your first report arrives the next morning.

  1. Connect your Git host

    GitHub, GitLab or Bitbucket. Merge requests, reviews and deployments come from here.

  2. Connect Jira

    Sprints, work in progress and flow come from your boards.

  3. Add monitoring and AI tools

    Google Cloud Monitoring or CloudWatch for error budgets. The AI tools you pay for, for cost and adoption.

  4. Add Slack and invite your teams

    The morning summary goes to Slack and by e-mail. The bot answers questions there too.

Work and code

  • Jira
  • GitLab
  • GitHub
  • Bitbucket
  • REST API

Where reports go

  • Slack
  • E-mail digest

Monitoring

  • Google Cloud Monitoring
  • CloudWatch

AI tools attributed

  • Claude Code
  • GitHub Copilot
  • Cursor
  • Codex
  • Gemini
  • Devin

Read the setup guides →

AtlassianAlso listed on the Atlassian Marketplace →

FAQ

Questions about AI Impact

How do you know a merge request is AI-assisted?

From the commits behind it. Claude Code, GitHub Copilot, Cursor, Codex, Gemini, Devin and Aider can add a Co-authored-by line to the commits they help write, and Agile Analytics reads those lines through your Git host’s API. Suggestions accepted while typing in the editor leave no such line, so they are not counted. The page also shows what share of merge requests it could read, so you know how far to trust the figure.

Which Git hosts does it work with?

GitLab and Bitbucket. Merge requests on GitHub are not attributed to AI tools at the moment; the rest of Agile Analytics works with GitHub as usual.

Is the comparison fair?

As fair as data from real work gets. Speed is compared inside each repository, never across them, and groups of fewer than five merge requests are left out. It is still a comparison, not a controlled experiment: a difference tells you where to look, not what caused it.

Does it rank individual developers?

Two pages do, and neither scores performance. On Enterprise, the AI leaderboard shows who merges with AI and how much, and an admin can switch it off. Kudos has a leaderboard of the recognition colleagues give each other. Health scores and reports stay per team, and what vendors bill per person is never shown per person.

Which plan includes AI Impact?

Scale-Up and Enterprise. Enterprise adds spend sync, which reads what Anthropic, OpenAI, Google Gemini, GitHub Copilot and Cursor billed every night, and the AI leaderboard. The 60-day trial includes every feature. See pricing.

Where is my data processed?

In the EU, under a GDPR data processing agreement. The privacy statement covers the rest.

Find out what AI delivers for your teams.

60 days free, every feature included. No credit card.

Or start your 60-day free trial →