Speed Is Easy, Trust Is Hard: AI's Real Impact on Developer Productivity
summary
AI boosts hands‑on coding efficiency — 40% more PRs per hour — yet org throughput rises only 7.8% overall PR velocity because coding is ~14% of time and just ~10% is available for new innovation. Telemetry from 60k+ engineers shows collaboration, security/compliance and ops eat capacity; 77% of devs rework AI code. Tool wins come with process fixes: Azure SR agent cuts human incident mitigation time 2.4x, and calendar hygiene for 3,000 engineers adds two hours/week and yields a 13% PR velocity lift. Track AI adoption and token cost separately from quality outcomes.
AI-written summary · about
chapters (9)
- 0:00Introduction and background
- 1:24Developer sentiment on AI productivity
- 5:41Why AI gains are smaller than expected
- 15:38Time constraints and bottlenecks
- 22:19AI adoption and maturity stages
- 30:28Overcoming AI tool adoption challenges
- 33:12Metrics and measurement pitfalls
- 36:43Future of AI in software engineering
- 41:44Closing thoughts and Q&A
related talks
18mHumans + AI: The Future of Software Development Teams
23mThe AI productivity paradox: Why feeling faster isn't the same as going faster
22mThe AI Productivity Paradox in Engineering
45mAgentic AI Coding Practices for Speed with Quality
27mThe state of AI in software development: Insights across 400+ organizations
19mThe AI Code Review Bottleneck: Escaping the Velocity Trap
29mLeadership in AI-assisted engineering
18mDoes AI Actually Boost Developer Productivity? (100k Devs Study)