we went from a small team to 20+ and release QA scaled worst. more devs shipping more Al-generated code meant more flows to re-verify before each release. it ate real hours. my weekends too. adding people didn't help. the verify work grew faster than the team.

we put the verify step inside the agent loop using testsprite CLI. the agent runs a real session against the live app. when something breaks it gets a failure bundle it can act on and rerun. cloud account was a mild pain to sort out. passing checks pile up into a suite that sticks around so coverage builds as we ship.

people still own the judgment calls and the fuzzy edge cases. moving the routine verify work into the loop stopped release QA from tracking headcount. how are other founders keeping QA cost flat while the team grows?