TZC-AIR measures what it would cost to rebuild this software today, with AI-assisted development — from the code itself, at any point in a deal, from first look to exit.
What would it cost a competent competitor — or the target itself — to rebuild this software from scratch today, with AI-assisted development? No standard technical due diligence resolves it. TZC-AIR does: a defensible effort-and-cost estimate with its confidence band, measured from the code.
Metadata leaves the premises — your source does not. Three independent estimates are reconciled into one figure with a band; where they disagree, the band widens and the report says so.
A scanner runs on the target’s premises and uploads derived metadata, with identifiers hashed. Your source tree is not copied out.[1]
Three independent estimators — complexity-weighted code volume on a Brooks/COCOMO lineage, effort inferred from development-history patterns, and the semantic structure of the code — run separately, then reconciled with outliers penalised and their divergence turned into the confidence band.
The AI-productivity assumption is printed on every report, with how the result moves if you assume differently. Calibration is versioned; every change is logged.
Third-party code, generated and build output, documentation and data files are excluded from the rebuild scope — by location, by content type and by lookup against a knowledge base of 30+ billion file hashes. The excluded share is shown.
The rebuild estimate as a band in person-months and euros or US dollars, the assumptions it rests on, and the risks that sit next to the number:
TZC-AIR measures one thing precisely: what it would cost to rebuild the technology. It is not a market valuation, and not a guarantee of business success. It prices the code — not the customer relationships, the brand or the market position that turn working software into a business.
One scan on a developer machine with Git access. No change to the deal timeline.
Built on hundreds of technical due diligences, two scanners with more than ten years in production, and a code knowledge base of 30+ billion file hashes.
[1] The exact content of the transferred metadata is open to your inspection throughout the engagement.
