Ask HN: How are PMs keeping up with AI-accelerated engineering output?

2 points by rudolftheone 12 days ago | 1 comment

With tools like Claude Code, Cursor, or Codex, engineers are shipping code faster than before. The bottleneck is no longer "how fast can we build it" but "how fast can we spec it well enough to build the right thing."

My Product Team is struggling to keep up: writing detailed, high-quality specs used to be fine when engineering was the constraint, but now I find them to think on rushing specs (and paying for it in rework) or becoming the drag on delivery.

Curious how other organizations are handling this:

- Have you changed how specs are written or structured?

- Are PMs using AI to write specs faster, and does that actually help or just move the problem?

- Have you restructured the PM-to-engineer ratio?

- Or have you accepted that "good enough" specs are fine when iteration is cheap?

eterm 12 days ago |

If your bottleneck is product spec rather than QA & testing, then you're doing well.

And that hints at one solution, if you demand better quality then you'll slow down engineering back to a level you can control.