01
The bill keeps growing
Per-developer licences, API usage, several tools at once: the AI budget line goes up every quarter.
Free diagnostic · 48 hours · Read-onlyCopilot, Cursor, Claude Code:
Copilot, Cursor, Claude Code:
what are they really earning you?
idweo measures the real impact of AI on your engineering teams — share of generated code, time saved, effect on quality — from your Git history. Team-level results, never individual monitoring.
48h
To receive your free diagnostic
€0
For the diagnostic, no commitment
0
Lines of code kept after analysis
Team
The level of detail in our results
The problem
You pay for AI every month. Nobody knows what it earns.
AI coding tools are now part of how teams work. The question from leadership always follows: is it worth what it costs?
02
Vendors measure usage, not impact
Accepted suggestions and active users don’t tell you whether your team ships more, or whether quality holds.
03
Leadership wants a number
Before renewing or expanding licences, your CEO or CFO asks for a return on investment. How the team feels about it is not enough.
04
The quality risk is invisible
More generated code can also mean more rework and incidents. Without measurement, you find out too late.
What you get
Five answers in numbers to decide on facts
Every commit is analysed: estimated effort, quality, share of AI-written code. We aggregate those measures by team and by month, before and after your tools were rolled out.
Share of AI code
How much of your code is AI-generated, by team, repository and month, and how it has moved since the tools were rolled out.
Time saved
The estimated effort behind each commit, compared before and after adoption, converted into hours and then euros.
Effect on quality
Commit quality and change failure rate, set against the share of AI code. So you know whether speed is paid for later.
Adoption by team
Which teams get value from AI and which don’t use it yet, so training goes where it is needed.
Cost against gain
What your licences cost against the value of the time saved, with a formula your CFO can check and adjust.
One page for leadership
Every report opens with a one-page summary a CEO or CFO can read in five minutes, with our recommendations on licences.
Who it is for
For companies with 15 to 150 developers
You rolled out AI coding tools in the last twelve months, and now someone is asking what they earn.
CTOs and engineering leaders
You own the AI budget.
Justify the spend with numbers from your own code, find the teams where AI really helps, and spot quality effects early.
Request the diagnosticCEOs and CFOs
You sign the renewals.
Get a clear number before expanding or cutting licences, with an explicit method you can challenge.
See pricingFractional CTOs and AI trainers
You work with several companies.
Show your clients the effect of your advice or training, before and after. 20% commission on the audits you bring.
Become a partnerHow it works
Four steps, without disturbing your teams
No agent to install, no interviews with developers. Read-only access to the Git history is enough.
1
A 30-minute call
We agree the scope: which repository, which AI tools, since when. A confidentiality agreement if you want one.
2
Read-only access
A read-only GitHub or GitLab token on the chosen repository. You can revoke it at any time.
3
Analysis in 48 hours
Every commit is analysed: effort, quality, share of AI code. Diffs are analysed and then deleted; only metadata and scores are kept.
4
A debrief with the founder
30 minutes to walk through the results, answer your questions and decide whether going further is useful.
Our commitments
What we don’t do
Measuring a team’s work takes trust. These are the limits we set ourselves.
We don’t keep your code
Diffs are analysed by a language model and then deleted. We keep only metadata (dates, volumes, files) and scores.
We don’t rank your developers
Our reports show team-level results. Author names and emails are anonymised before the language model sees anything.
We don’t resell anything
Your data is never shared or sold, and it is deleted on request at the end of the engagement.
We don’t lock you in
The diagnostic is free, the audit is a one-off, and the monthly plan has no minimum term.
Free · No commitment
Start with the free diagnostic
One repository, six months of history, 48 hours. See what AI is already changing in your team, without paying or installing anything.
Read-only access, revocable at any time. Confidentiality agreement on request.
- Share of AI code in the repository, month by month
- How effort and quality have moved since your tools were rolled out
- A 30-minute debrief with the founder to read the results together
Pricing
Simple prices, known upfront
Set by the size of the engineering team. Prices exclude VAT.
Diagnostic
Free
1 repository · 6 months of history
AI ROI audit
€2,900
Up to 30 developers
AI ROI audit
€4,900
31 to 100 developers
Monthly tracking
from €390
per month · up to 20 developers
Straight answers
The questions we get asked
We are early, and we would rather answer the hard questions here than leave you guessing.
“You are new. Why should I trust you?”
Fair question. The method was built and validated on a real engineering team. The diagnostic is free so you can judge the work itself, and the first three audits are half price in exchange for a testimonial.
“How do you detect AI-written code?”
A language model reads the code each commit adds and estimates how likely it is to be AI-written, against explicit criteria: naming, comment style, boilerplate, error handling, structure. Signals measured automatically on the same code set the confidence score. The result is read as a trend by team and by month, never as a verdict on a single commit.
“What exactly do you see of our code?”
The diffs of commits in the repositories you open, read-only. They are analysed by a language model and then deleted; we keep metadata and scores. If your policy requires it, the analysis can run with your own key or a model you host.
“Is it legal to measure developers’ work in France?”
Our reports are team-level, with no individual ranking. For continuous tracking, French labour law requires informing employees and consulting the works council (CSE) before deploying this kind of tool: we help you prepare that.
“Why not GitHub Copilot’s or Cursor’s own dashboards?”
They measure usage of their own tool: accepted suggestions, active users. idweo measures the outcome across all tools: what was delivered, with what effort and what quality.
Free · Read-only · 48 hours
Measure before you renew your licences
The free diagnostic is the fastest way to find out whether AI is already paying off in your team.
Or book 30 minutes to talk it through: we will tell you straight if idweo is not right for you.