Product management
Thoughts on Marty Cagan's article on 'AI and the Future of the Product Role'
Where Cagan gets it right, where I'd take it further, and why I think the role is heading from product manager to product operator.
Marty Cagan needs no introduction. Inspired was instrumental in me becoming a PM. When I was moving from business development and marketing into product, the ideas in that book taught me how to think about building products, and they’ve stayed with me ever since.
So when he published AI and the Future of the Product Role, first as his keynote at SVPG’s 2026 Product Leader Summit and then as an article, I read it closely. It’s created quite a buzz. Here’s my take, from ten years in product management, working in both the project model and the product model he describes.
What Cagan argues
- The operating model decides the role. Companies on the project model have feature team PMs and product owners. Companies on the product model have empowered team PMs.
- AI is taking over delivery. As agentic tools absorb delivery work, fewer feature team PMs will be needed, and the dedicated product owner role is being phased out.
- Discovery is the job. Once delivery stops being the bottleneck, the core work is discovering what’s worth building. That depends on product sense and judgment, which are hard to automate.
- Empowered PMs matter more on AI products. Probabilistic products put the PM at the centre of evals and risk, so demand for empowered PMs will keep rising.
- Most companies won’t change. Many will never leave the project model, so product owners should upskill fast.
Where he gets it right
Outcomes over outputs. This doesn’t change with AI. If anything, when building gets cheaper, shipping things matters even less on its own. The question is still whether what you shipped did its job.
Discovery is the focus. I made the same point in my earlier post on what isn’t changing in product management with AI. When AI makes building cheaper, the parts it doesn’t touch become more important, not less: picking the right problem, deciding with incomplete information and checking whether it worked. As I put it there, “When anyone can ship a feature in a week, the advantage goes to the teams that know what to build next.”
Product owners need to upskill. I’ll admit my bias here. I’ve never been convinced by product ownership as a separate role. A user-centric engineer with good communication skills could do the job equally well, if not better. So I do see the role as being at risk, and I’d take Cagan’s advice seriously if it’s yours.
Taking it further
Cagan’s argument is a reasonable starting point. Here’s where I’d take it further.
Operating models will change, and I welcome it. Software factories and agency models will have to move upstream. When delivery is cheap, taking a spec and building it is no longer much of a business. Maybe, just maybe, that lifts the quality of products out there, especially in complex SaaS and enterprise software.
Upskilling has never been easier. Remember when it was hard to build anything and learn by doing? In the old days of PMing, a team simply had to trust its PM. Now it’s easy to produce a body of work: prototypes, hobby projects that validate an idea, even side hustles. AI has made that easy, and it’s the best answer to “how do I upskill?” that I know.
Product leaders need to be hands-on. Cagan leaves leaders out of the article, but I think they’re affected as much as anyone. How do you lead a product organisation if you don’t understand how the operating model and the role are changing with AI? I expect leaders to become much more hands-on, because you can’t direct something you haven’t tried.
My hot take: from product manager to product operator
Here’s how I see the role evolving:
Product owner → product manager → product builder → product operator
A product builder has the tools to build fast, whether that’s for discovery or for delivery. A product operator goes a step further and runs a swarm of agents (product, design, engineering, research, analytics) to take ideas all the way to production. The tooling and platforms are getting there. Venture capital firms have operators, so why not product?
A pinch of salt
I respect Cagan and what he’s done for product management. But posts like this are aimed at people who are new to the job. Product influencers mostly work as consultants: they advise companies and then share their views, without being present at every stage of the work. That shows. The article felt predictable and surface-level to me, and it gave few references for its bigger claims.
Maybe I’m just not the audience for it anymore.