David Pereira’s take on the fake PM problem landed in my inbox last week, and one line has stayed with me: “If you removed the roadmap and the Jira board, would you still know what to do?”
I’ve been asking myself a harder version of that question ever since. Not just for the PMs on my team — for myself.
The honest answer, in most cases, is no. And I don’t think that’s a personal failing. It’s a structural one. We’ve built PM practice around artefacts — PRDs, roadmaps, sprint plans, OKRs — and confused the artefacts with the thinking. The documents became proof of work. The process became the product.
Now AI can generate the artefacts faster than ever. And that should be the wake-up call it isn’t.
The Fake PM Pattern
Fake PMs — and I’m being deliberately provocative, not dismissive — manage artefacts. They maintain backlogs, write specifications, facilitate ceremonies, produce status reports. They’re busy, they’re competent, and they’re extremely good at looking like they’re doing product management.
What they’re not doing is making hard trade-offs under uncertainty. They’re not owning outcomes. They’re not accelerating and maximising the creation of value. They’re not deciding — they’re processing.
What happens when conditions change? When the roadmap becomes irrelevant? When the data contradicts the plan? When a stakeholder asks a question that isn’t covered by the last slide deck? In those moments, the artefact-managers stall. They need a process to follow. They need a document to reference.
Ben Yoskovitz put a version of this plainly: the PM role isn’t disappearing, but the gap between PMs who think and PMs who produce is about to become visible in a way it never was before. AI building is easy; figuring out what to build and getting it into the right hands is as hard as it’s ever been.
AI Is Accelerating the Wrong Version
Here’s the uncomfortable part. Every AI tool marketed at product managers is optimised for artefact production. Better PRDs. Faster user story generation. Automated sprint summaries. Cleaner OKR frameworks.
These tools are genuinely useful. I’m not arguing against them. But they make the surface activity easier without touching the underlying judgment. And in doing so, they create a very effective illusion of progress.
A fake PM with good AI tools is indistinguishable from a real PM — until the moment the situation requires a call that isn’t in the playbook. At that point, the tools don’t help. The output quality collapses. The gap becomes obvious.
The irony is that AI was supposed to free up cognitive bandwidth for higher-order thinking. In practice, for many PMs, it’s filling that bandwidth with more surface activity. Faster specs. More polished decks. Longer context-building documents. Zero improvement in decision quality. No increase in value creation.
What the Real Test Looks Like
Strip away the tools. No Jira. No Notion. No roadmap doc. No last quarter’s OKRs to reference.
Now: what does your product need to do over the next 90 days, and why? Who is the customer you’re building for, and what do they need that they can’t get elsewhere? Where are you making trade-offs, and on what basis?
If you can answer those questions clearly — not fluently, clearly — you’re a PM. If you hesitate, or if your answers are mostly process artefacts restated as strategy (“we’re focusing on retention this quarter per the company OKRs”), there’s a gap worth closing.
This isn’t about working without tools. Tools are necessary. It’s about whether the thinking exists independently of the tooling. Whether the judgment runs underneath the artefacts, or whether the artefacts are substituting for the judgment.
Three Things That Actually Build It
First: make your reasoning legible. Don’t just write the decision — write the reasoning behind it. Why this, not that. What you’d do if the data were different. This sounds obvious. Most PMs don’t do it.
Second: seek the moments where the playbook doesn’t apply. The situations that require judgment rather than process are the only training ground for judgment. If your week is entirely ceremony and documentation, you’re not getting the reps.
Third: own outcomes, not delivery. This is harder than it sounds, because most PM incentive structures reward delivery — shipped features, closed sprints, completed roadmap items. Real PM practice means defining what success looks like before you build, tracking it after you ship, and being willing to say “this didn’t work” when it didn’t.
AI is making the artefact side of product management faster, cheaper, and easier to fake. That’s not a reason to panic. It’s a reason to be clear about what the job actually is — and where the irreplaceable part lives.
The irreplaceable part isn’t the PRD. It’s the judgment that informs it.
This post was prompted by David Pereira’s “Most PMs Aren’t Real PMs (And AI is making things worse)” and Ben Yoskovitz’s “They Tried to Kill Product Managers. Now Everyone Needs to Be One.”
Leave a Reply