Measuring AI
on facts, not impressions
idweo helps companies find out what their AI coding tools really earn, from their own Git history.
In two years, AI coding tools have become part of a large number of engineering teams. Licences add up, and the question soon reaches the leadership team: what does it earn?
Vendor dashboards measure usage of their own tool. They don’t say whether the team ships more, or whether quality holds. Public studies contradict each other, and none of them is about your code.
idweo answers the question from the most reliable source you have: your commit history. The method was built and validated with a real engineering team before being offered to other companies.
How we work
Three principles behind every diagnostic and every audit
Our mission
Give CTOs and leadership teams a reliable number on AI’s impact, to decide where to invest, where to train and which licences to keep.
Our method
Every commit is analysed — effort, quality, share of AI code — then results are aggregated by team and by month. The formula is explicit and adjustable.
Our line
Team-level results, never individual rankings. Code analysed and then deleted. And we tell you straight when idweo is not right for you.
What we believe
The principles behind our reports
Usage is not impact
Accepted suggestions are not a shipped feature. We measure what lands in the code, not the number of clicks.
Speed without quality is expensive
More code in less time can also mean more rework. We always look at both together.
The team’s trust comes first
A measure seen as surveillance does not survive. That is why our results are team-level and explained openly.
A number must be checkable
Your CFO should be able to redo the calculation. Our assumptions (hourly cost, licence cost) are written down and adjustable.
Lionel Clemenson
Founder of idweo · Lyon
I present every diagnostic and every audit myself. If the numbers show nothing useful for you, I will tell you.
idweo is a brand of MEKANO SAS, registered with the Lyon trade register (RCS).
Judge us on your own numbers
The free diagnostic is the most honest way to evaluate us: 48 hours, one repository, no commitment.