The part that feels under-discussed is that AI may not eliminate PM as a role, but it absolutely raises the bar for what counts as real product work.
A lot of PM work was already splitting into two very different jobs. One is high-agency product leadership: picking the right problem, understanding users deeply, making tradeoffs. AI is brutal for the second category. If agents can write specs, summarize research, draft PRDs, generate prototypes, analyze feedback, and remember context across sessions, then the coordination-only PM gets exposed pretty fast.
That is why I think the āPM is the obvious dream jobā narrative is aging badly. I wrote a longer version of that argument here:
The part that feels under-discussed is that AI may not eliminate PM as a role, but it absolutely raises the bar for what counts as real product work.
A lot of PM work was already splitting into two very different jobs. One is high-agency product leadership: picking the right problem, understanding users deeply, making tradeoffs. AI is brutal for the second category. If agents can write specs, summarize research, draft PRDs, generate prototypes, analyze feedback, and remember context across sessions, then the coordination-only PM gets exposed pretty fast.
That is why I think the āPM is the obvious dream jobā narrative is aging badly. I wrote a longer version of that argument here:
https://consulting2tech.substack.com/p/2-please-dont-become-a-product-manager