Blog · topic

METR

1 article on METR.

All architecture AI tools AI development comparison engineering leadership productivity reliability vibe coding AI agents DORA AI productivity code comprehension code review procurement agent economics AI strategy alternatives benchmarks cost CTO governance hallucinations leadership no-code requirements risk startups AI context AI costs AI prompting AI ROI AI verification board reporting bus factor change impact compliance context engineering Copilot cross-functional Cursor data boundary developer experience developer productivity efficiency engineering decisions engineering economics engineering metrics engineering practice engineering risk evaluation GenAI hiring institutional knowledge integrations legacy migration LLM efficiency LLM tokens market analysis MCP methodology METR metrics neuro-symbolic AI on-premise AI onboarding organisational cost pricing research security semantic distance strategy teams technical debt token optimization vendor evaluation visual architecture

The 19% Problem: What a Controlled Trial Revealed About AI and Developers

August 25, 2026

The 19% Problem: What a Controlled Trial Revealed About AI and Developers

A randomised controlled trial found experienced developers were 19% slower with AI — while estimating they were 20% faster. The detail everyone skips explains why.

AI productivityMETRDORA