Gergely Orosz

Gergely Orosz writes The Pragmatic Engineer, and his reporting on AI coding tools follows one pattern: interviews with the people who built the thing, then the numbers they will disclose. In the article Claude Code (15 September 2025) he spoke with founding engineers Boris Cherny and Sid Bidasaria and product manager Cat Wu, and reported that roughly 90% of Claude Code is written by Claude Code itself, and that pull requests at Anthropic rose 67% while the engineering team doubled.

The article Why Ramp built its own in-house coding agent, Inspect (25 August 2026) applies the same method to a fintech that chose to build rather than buy: Rahul Sengottuvelu, Hamid Dadkhah and Zach Bruggeman describe an agent that now accounts for 75% of merged pull requests at Ramp, against about 60% in January. Orosz holds that the "buy, don't build" convention may not apply to AI tooling, and Ramp is the case he puts behind it.

With Laura Tacho, CTO of DX, he published the AI Measurement Framework after examining how 18 companies, among them Google, GitHub, Microsoft and Dropbox, value their AI spend. The finding that organizes it: 85% of engineers use AI tools, and engineering leaders still cannot say what that is worth. Suggestion acceptance rate, once the reference metric, is losing credibility because it captures neither maintainability nor introduced bugs. Orosz expects agent telemetry to change substantially.