AI review pushed pre-release catch rate from 64% to 86% — and no QA jobs were cut
alex_verem · x · 2026-09-30
Responding to 'will agentic AI replace QA?', mardehaym shares a client case: turning on automated AI review across all PRs boosted dev throughput overnight but left releases piling up. The team rebuilt QA around risk — automating regression on core systems and splitting release checklists into what scripts catch vs. what needs human judgment. Every PR now clears a verify-understand-explain gate before human review, lifting pre-production catch rate from 64% to 86%, with 22 releases and zero rollbacks in Q1 and zero QA layoffs. AI changed what QA could accomplish, not whether the function exists.
More from coding & agent
- Monid, 'OpenRouter for Agent Tools,' Unifies 2,000+ Tools Across 72+ Providers — aigclink · 2026-09-30
- Monid Goes Open Source: Declarative Connectors Give Agents 2,000+ Tools via One Key — aigclink · 2026-09-30
- Claude-Powered SEO Content Pipeline Goes Viral — Then an SEO Vet Shares What Happens 3 Months Later — lilyraynyc · 2026-09-30
- NVIDIA open-sources ProRL Agent: rollout-as-a-service to fix RL training bottlenecks for LLM agents — burkov · 2026-09-30
- 100 autonomous AI agents are building a civilization in open-source sim CYMONIA — Positive-Captain-709 · 2026-09-30
- How a runaway agent loop almost burned $1,200 overnight, and the case for hard budget caps — MaverikSh · 2026-09-30