SECOND OFFER · AI SPEND AUDIT · BOOKING NOW
A three-week audit of your Copilot, Cursor and Claude spend: real usage per seat, code attribution in your repos, and the number your CFO is asking for at renewal.
THE OFFER
Renewal conversations run on vendor dashboards, and vendor dashboards report activation rather than use. This measures the other side of it, from your own systems. How we handle your data →
No platform licence, no per-seat fee, and no agent installed to collect the measurement. Everything is read from vendor admin consoles you already pay for and repositories you already own.
WHAT YOU GET
Delivered as files you own, in formats engineering and finance both already read.
01
Active versus licensed, per seat, for every Copilot, Cursor and Claude subscription in the estate — including the seats nobody has opened since onboarding.
02
Volume by team, measured from vendor attribution markers and commit trailers already present in your repositories. No agent is installed to collect this.
03
The real denominator: spend divided by people actually using the tool, by team and by vendor.
04
Which contracts to grow, hold, cut, or consolidate at the next renewal, with the seat counts behind each call.
05
Delivered to engineering and finance together, so the number survives the room it is questioned in.
HOW IT RUNS
Three weeks is long enough to measure a real contribution window and short enough to land before the renewal date it is meant to inform.
Read-only access to vendor admin consoles and your Git host. We establish the licensed-seat baseline and the repositories in scope.
READ-ONLY THROUGHOUTWe measure AI-attributed contribution across the scoped repositories and reconcile it against seat activity.
READ-ONLY THROUGHOUTFindings, cost per active seat, and a renewal recommendation — delivered as files you own, read out in sixty minutes.
READ-ONLY THROUGHOUTWHO IT'S FOR
VP Engineering, Head of Platform, or the finance partner who owns the tooling budget, at companies with 200–3,000 engineers.
YOU ARE ASKED
What we are getting for the AI tooling line, in a budget review, and the honest answer today is a vendor-supplied activation number.
YOU SUSPECT
That a meaningful share of paid seats are inactive, and that the teams generating the most AI-attributed code are not the teams with the most licences.
YOU NEED
A defensible per-seat number before the renewal conversation, produced by someone who does not sell the seats.
WHAT THIS IS NOT
This measures usage and attribution, not developer performance. It does not rank engineers, it does not produce per-person output metrics, and it is not delivered in a form that can be used for performance management. If that is what you want from it, we are the wrong vendor and will say so on the call.
THE OTHER ENGAGEMENT
The AI Spend Audit asks what the seats cost and what they produced. Agent Flight Check asks what the agents did and what they were allowed to touch. Companies usually discover they needed both, in that order, after an incident. Run together they are scoped as one engagement — ask on the call.
THREE WEEKS · $9,500 · READ-ONLY THROUGHOUT
Thirty minutes to confirm scope and dates. If the numbers you need are already in a dashboard you own, we'll tell you that instead of selling you an audit.
Book the readout call →