Doctolib Rebuilt Its Testing Infra in Hours, Not Weeks — and Ships 40% Faster
Anthropic's 2026 enterprise report has a case study most people read as a testing story. It's actually an infrastructure story: the agent's real win was replacing legacy test scaffolding, not writing tests.
NeuroX AI · July 14, 2026
Buried in Anthropic's 2026 enterprise report is a line that reads like a testing win but is really an infrastructure one. Doctolib used Claude to replace legacy testing infrastructure in hours instead of weeks — and now ships features 40% faster.
Read that again. The headline number isn't "the agent writes our tests." It's "the agent rebuilt the scaffolding those tests run on." Legacy test infra is where most teams' velocity quietly dies: flaky harnesses, slow suites, fixtures nobody understands, a setup step that takes a new hire a week to get green. That's not a model problem. It's the exact production layer a prototype never has to think about.
The 40% comes from removing a bottleneck, not from typing faster. When the harness is fast and trustworthy, every downstream change ships with confidence instead of a manual QA pass. Compress weeks of infra migration into hours and you don't just save the weeks — you unblock every feature that was waiting behind them.
This is the pattern we see over and over: the demo proves the model can code. Production is decided by the plumbing around it — test infra, CI, the seams where a change either ships safely or doesn't.