Every ROI number gets litigated. This one is built to survive it.
A net dollar figure and a return multiple, built bottom-up from real sessions, net of the AI Tax.
After AI spend, and after the impact lost to rework.
From an estimate, to a number built out of the org’s own sessions.
- A vendor ROI calculator with your headcount typed into it
- Seat spend on one side, a feeling of progress on the other
- A number the CFO can question and nobody can defend
- Net effective gain in dollars, and a return multiple
- Value created and AI spend both attributed down to the team
- Every line traceable to the sessions and merged PRs behind it
How the number is built.
Four steps, in order, each one you can defend.
Cycle-time savings
per merged PR
Loaded engineering cost
your actual cost
AI Tax
rework + review drag
Total AI spend
licence + sessions
Operational AI ROI
net gain + multiple
Return, and the places it leaks away.
From a single team to the company case.
A team's return, attributed and ranked, then rolled up to the org.
ROI tells you the return. Drivers tell you the levers.
Or explore the rest of the platform
Questions leaders ask.
Cycle-time savings per merged PR, measured against the same team's non-AI baseline and valued at your loaded engineering cost, minus total AI spend for the period. Impact lost to rework and review drag is deducted before the net figure, which is why it reads lower than a gross savings number.
No, and the page says so. This is operational AI ROI: the return on the coding workflow, the part Tetriz can build from sessions and shipped code. It's a subset of the engineering org's total AI return, not the whole of it.
Vendor calculators start from seats and assumptions you supply. This one is built bottom-up from your own sessions and merged PRs, so each line traces back to the work it came from rather than an input you typed in.
It's the gap between what AI costs and what it actually ships: review burden, rework, and code that never lands clean. It comes off before the net figure, which is why this number indexes on net rather than gross.
The share of AI spend that went into sessions which didn't convert into shipped output. It's the headroom in the number, and it's usually a harness problem you can fix.