Updated
Updated · InfoWorld · Oct 7
Author Shifts to AI-Directed Coding After Cursor Fixes 20+ Threading Bugs
Updated
Updated · InfoWorld · Oct 7

Author Shifts to AI-Directed Coding After Cursor Fixes 20+ Threading Bugs

1 articles · Updated · InfoWorld · Oct 7

Summary

  • More than 20 race-condition scenarios in the author’s own code were flagged and fixed by Cursor within minutes, prompting him to say he no longer really writes software himself.
  • Cursor is already built into his workplace pull-request review flow, where it scans submitted code for problems that humans often miss—especially in complex multithreading.
  • The author says the change happened in just a few months: he now writes only a few lines here and there and mainly directs tools such as Cursor, Claude Code and Codex.
  • David Heinemeier Hansson recently urged developers to “put down our pencils,” and the author argues coding agents could also make broader product work—such as native apps across platforms—more feasible.

Insights

With AI generating half of all code today, who is actually catching the hidden bugs slipping past overwhelmed human reviewers?
If most developers distrust AI-generated code, why are tech teams allowing autonomous agents to write the vast majority of their software?