The AI coding tools market now has at least five serious competitors - Cursor, GitHub Copilot, Aider, Windsurf, Amazon Q Developer - each with their own benchmark claims and demo videos. Scott Hanselman joined Software Engineering Radio in March 2026 to offer something rarer: a working developer's honest account of where these tools actually help and where they fall short.
The episode (SE Radio 711) covers the practical questions that don't get enough coverage: which tools are reliable on production code rather than toy projects, where AI coding assistants fail quietly by writing confident-sounding wrong code, and how to build a workflow that treats AI as a tool rather than an authority.
Hanselman spent years at Microsoft as a developer advocate, which gives him an unusual combination - deep technical credibility and a genuine ability to explain complex ideas plainly. His perspective cuts through promotional noise because he's used these tools under real conditions, not just for conference demos.
If you're trying to decide whether Cursor, Copilot, or something else deserves a serious trial on your actual work, this episode is worth an hour of your time.