David Crawshaw (crawshaw.io)
- title
- David Crawshaw (crawshaw.io)
- type
- entity
- summary
- David Crawshaw's blog at crawshaw.io — co-founder of exe.dev, ex-Tailscale, ex-Google; writes on agent-driven engineering, small teams, and infrastructure
- tags
- blog, agentic-coding
- created
- 2026-05-10
- updated
- 2026-07-21
David Crawshaw runs crawshaw.io. He's a co-founder of exe.dev (a nine-person team using agents heavily for product work), an ex-Tailscale engineer, and earlier worked at Google. His writing focuses on what changes when agents become useful collaborators — small-team dynamics, the breakdown of corporate review processes, and the operational machinery that has to compensate.
The May 2026 post agent-principal-agent-problem is one of the cleanest articulations of why agents make code review fail: it's not workload, it's the collapse of the effort signal that low-trust review depended on. The conclusion ("until someone develops processes that scale agent use in low-trust environments, small teams have a force multiplier big teams don't have, ship while you can") is the kind of thing that lands harder if you've spent time at a BigCo trying to merge anything.
Not to be confused with blog.exe.dev, the company blog. Crawshaw's site is personal and diagnostic; the company blog runs to worked practice reports, and claude-is-not-a-compiler (Josh Bleecher Snyder, July 2026) is effectively the constructive answer to the problem agent-principal-agent-problem posed here.
Ingested articles
- agent-principal-agent-problem — 2026-05 — Why agents broke code review by collapsing the effort signal between contributor and reviewer