Find out what your AI-written code is actually costing you in review, rework and incidents.
Cursorship compares AI-assisted changes with human-authored ones in the same codebase on review time, rework, defects and incidents, for engineering leaders. It does not rank individual developers.
No account. No card. No sales call.
- AI-assisted and human code compared
- Never ranks individual developers
- Self-hosted analysis available
Does AI-assisted code save more time than it costs to review?
Put in your team, how much of its code is AI-assisted, and what those changes save to write and cost to review and rework. See the net hours and cost each month. Runs in your browser. No account, no card, no call.
How it adds up, each month
- AI-assisted changes a month
- 347
- Writing hours saved a month
- 173 hours
- Extra review hours a month
- 69 hours
- Rework hours a month
- 69 hours
Take this with you
No email required. It is your result.
Cursorship — AI code net-time calculator · 40 developers, 40% of changes AI-assisted · Net 35 hours a month after review and rework · 80% of the writing time saved goes back into review and rework Run it yourself: https://cursorship.com
Method: assisted changes = developers x changes a week x assisted share, over a year and divided by months. Net hours = writing time saved - extra review - rework. Every input is your own estimate, so the result is only as good as they are; measuring these from your repository is what the AI Code Impact Report does.
What happens after the code is written
Measurement of downstream consequences rather than adoption, compared within the same codebase and team.
- Provenance linked to outcomes
- Code provenance is linked to what happens to that code afterwards.
- Review time and comments
- Review time and comment volume for AI-assisted changes against human-authored ones.
- Reverts and rework
- Revert and rework rates, compared side by side.
- Defects and incidents
- Defect density and incident linkage.
- Time to production
- Time from first commit to production.
- No individual rankings
- It does not rank individual developers, and that is stated as policy.
The problem
AI assistants write a large share of code now, and organisations have no idea what that means.
Some teams ship faster, others have quietly traded authoring time for review time, rework and incidents, and leadership is making headcount and tooling decisions on anecdote.
The vendors report acceptance rates, which measure nothing anyone cares about.
Who it is for
VPs of Engineering and CTOs
Leaders at 30 to 1,000 person companies where developers use AI coding assistants heavily and nobody can say whether it is helping.
Platform leads
Leads who need to know where AI assistance speeds work up and where it adds review time and rework.
How it works
Cursorship, from the first step to the result.
- 1
Connect a repository read-only
For the free AI Code Impact Report.
- 2
See six months of history
How much code shows assistance patterns, and how it compares with the rest.
- 3
Measure the consequences
Review, rework, defects, incidents and time to production, within each codebase and team.
- 4
Find where it helps
And where it produces code that costs more to review than it saved to write.
Questions people actually ask
How is this different from acceptance rates?
Acceptance rates measure adoption. Cursorship measures what happens afterwards: review time, rework, reverts, defects, incidents and time to production.
How does it compare AI-assisted code?
It links code provenance to downstream outcomes and compares AI-assisted changes with human-authored ones within the same codebase and team.
Does it rank developers?
No. It does not rank individual developers, and that is stated as policy.
What is in the free report?
Connect a repository read-only and get an analysis of the last six months: how much code shows assistance patterns, and how that code compares with the rest on review time, rework rate and defect linkage.
Can the analysis run without sending code externally?
Yes. Self-hosted analysis comes with the pricing above 250 developers, for companies that will not send code externally.
What does it cost?
The report is free. After that it is 12 dollars per developer per month, or 8 dollars per developer above 250 developers, with self-hosted analysis.
Is there an enterprise plan?
Yes, with an SLA and data-residency controls.
Why should I trust the AI code net-time calculator?
It is arithmetic on the estimates you enter, worked out in your browser. It does not read your code; the free AI Code Impact Report does that from your repository's history.
Pricing
The impact report is free. Ongoing measurement is priced per developer.
Impact report
The AI Code Impact Report from your repository's history.
- Read-only repository connection
- Review, rework and defect comparison
Team
Ongoing measurement of AI-assisted code.
- Review and rework metrics
- Defect and incident linkage
- Time to production
Scale
For companies that will not send code externally.
- Self-hosted analysis
Prices in USD.
Does AI-assisted code save more time than it costs to review?
Put in your team, how much of its code is AI-assisted, and what those changes save to write and cost to review and rework. See the net hours and cost each month. Runs in your browser. No account, no card, no call.
Open the free toolIt runs in your browser. Cursorship never sees your inputs.