The Pulse: Details of incident.io’s AI spend
incident.io reports that AI agents now generate code for nearly 80% of production pull requests, driving a 6× rise in “effort points” and a 60% drop in incidents per 1,000 PRs. The company’s AI spend has climbed to roughly $1.6 M this year (~$2,700 per engineer per month), while release velocity rose 47% to 2.8 releases per engineer per day. These numbers illustrate how large‑scale agent integration can dramatically boost developer throughput and reliability, but also requires substantial tooling investment.
⚡ Key Takeaways
- 80% of PRs shipped to production now contain code generated by an AI agent.
- Claude Agents and worktree tooling were adopted in June 2025 to enable parallel agent work on the same codebase.
- PR size grew 4.8× (22 → 111 lines) and PRs with no human reviewer rose 3.6× (12% → 43%).
- AI spend is on track at $1.6 M annually, equating to $2,700 per engineer per month.
- Releases per engineer per day increased 47% to 2.8 releases/eng/day (up from 1.9 in 2024).
Engineers can expect that scaling AI agents across the codebase can slash incident rates and accelerate release cadence, but must budget for high per‑engineer tooling costs and ensure robust parallel‑workflow support.
✅ Practical Steps
- Deploy Claude Agents alongside worktree tooling to enable multiple agents to operate concurrently on shared repositories.
- Track “effort points” using an AI model that ranks PR complexity from XS to XL to monitor productivity gains.
- Allocate budget for AI tooling at approximately $2,700 per engineer per month to sustain current velocity.
Want the full story? Read the original article.
Read on Pragmatic Engineer ↗