The AI Velocity Trap: Shipping Faster Without Breaking More
summary
Claim: AI-assisted velocity often shifts defects downstream and raises regressions and incidents. Evidence: agents (Code Rabbit, Co-pilot review, Cursor review, Codex, background agents) are used to shift review left, run QA agents that drive deployed URL like a real user, produce screenshots or video replay as proof, and run a 72-hour production watch. Practical system change: capture intent as first-class artifact and wrap every stage in zero-trust harnesses that verify intent against diffs, tests, and observability.
AI-written summary · about
chapters (6)
- 0:00AI velocity trap intro
- 2:55Faster code vs better software
- 7:05Intent as core development
- 13:06Zero trust workflow harnesses
- 20:07Monitoring intent in production
- 25:10Recap AI-driven principles
related talks
34mAI Agents for Performance: Ship Faster, Pay Less
22mBeyond Bugs: How AI turns QA into Full-Circle Engineers
26mFrom autocomplete to agents: AI coding assistance state of play
27mAI-First Software Maintenance & AI-First Software Delivery
30mArchitecting Integrity: Responsible Software Evolution in the AI Era
47mAI, Automation & DevEx: Fueling High-Velocity Engineering
31mThe Copilot Wrote It but Who Saves It: Surviving the AI Code Tsunami
35mBeyond the Hype and Hyperbole