How AI Is Changing Product Management: What Employers Now Ask PMs to Do

Most commentary on artificial intelligence and product management asks how many AI product jobs exist. This paper asks a different question: what does the content of the product manager (PM) job look like when the product is AI? Using 1,082 unique PM job postings at 165 companies, harvested on 9 November 2026 from the public Greenhouse, Ashby and Lever job-board APIs, I compare the 496 postings classed as AI-focused with the 586 that are not, treating the AI-focused set as a leading indicator of where the role is heading. AI-focused postings mention hands-on prototyping or coding about twice as often (36.9% vs 17.9%), a technical background more often (40.7% vs 30.2%), go-to-market work more often (42.3% vs 25.4%), and evaluation of models far more often (13.7% vs 1.0%). The classic core of the job does not move: roadmap ownership, stakeholder work, written communication and strategy appear at statistically indistinguishable rates in both sets. The prototyping, evaluation and go-to-market gaps survive a within-company comparison across 61 firms that post both kinds of role. I organise the findings into a framework in which the AI-era PM adds three roles, builder, evaluator and deployer, on top of an unchanged core; 15.7% of AI-focused postings ask for all three, against 6.0% of others. A contrast within one interview-practice question bank (695 questions derived from real job descriptions versus 3,427 classic questions) suggests that preparation material written around the classic core under-represents the new roles: 52% of JD-derived questions touch AI, agents or evals, against 5% of classic ones, while estimation questions fall from 6% to 0%. Data: AllthingsPM PM Hiring Index, November 2026 (1,082 Product Manager job postings, 165 companies), doi:10.5281/zenodo.24072616. Part of the AllthingsPM PM Hiring Index working paper series. Author: Arpit Ranjan, AllthingsPM (allthingspm.app). Conflict of interest: the author is the founder of AllthingsPM, an AI product management course and interview-practice platform mentioned in this paper.