Product managers today operate in a landscape of abundance—abundant data, abundant opportunities, abundant noise. Everyone has an idea. Every PM wants “just one more” experiment. The surge of generative AI has only amplified this frenzy. Yet, the job of a great product manager—especially in this era—is not to chase every one of them. It’s to know which ones actually needs tending.
The Flood of “Ideas”
Every team—from sales to operations to customer care—has a wishlist. Some of those ideas stem from real customer pain. Many are driven by internal incentives. Add AI to the mix, and now everyone wants an LLM-powered chatbot, a co-pilot for their teams, a real-time trend predictor, and a generative campaign builder by next quarter.
The PM’s job is not to say yes to every idea. It’s to interrogate them.
What problem does this solve?
Whose problem is it, really?
What does success look like—and is that success even measurable?
Too many roadmaps crumble under the weight of “important-sounding problems.” The result? You deliver features that look innovative, but don’t move the business. Or worse, features that move nothing at all.
Generative AI Is Not the Strategy
It’s tempting to treat generative AI as the strategy. But it’s just a tool. The real question is: What is your business trying to become? You don’t need a GPT wrapper on every customer touchpoint just because you can build one.
A product manager working in AI should start not with the model, but with the moment.
What are the moments in a customer’s journey where uncertainty blocks action?
What decisions does a partner team member make a hundred times a day that software can simplify?
What context does an algorithm lack that a generative layer could fill in?
Start there. Because the moment is where impact lives.
The Discipline of “No”
Saying “yes” is cheap. Saying “no” is costly. But that’s where product judgment lives. You’ll be asked to support a prototype that has an executive sponsor. You’ll get pressure to prioritize “that one thing Legal really wants.” You’ll face suggestions that “won’t take much engineering time.” It is your job to say no when the idea doesn’t tie to a customer journey, a business driver, or a measurable outcome.
A good product manager filters. A great one teaches others to filter.
Most importantly, Ask Better Questions
Before greenlighting any project, ask:
Are we solving a priority problem? Not a theoretical one, but one felt by customers.
Do we have the foundation? AI is expensive, both in infra and failure. Do we have the data, the integration points, and the organizational will?
Is it measurable? Can we isolate the impact? Will we even know if it worked?
Does it scale? If the pilot works, can we roll it out across categories, platforms—or will it die in a corner?
If the answer to most of these is no, don’t do it. At least, not now.
Zoom Out, Then Zoom In
Generative AI has opened up a new design space for everyone. We can reimagine everything from how customers interact with our business across every thoughpoint. But don’t start with the tech. Start with the value.
Zoom out. What is the business trying to become in 3 years?
Zoom in. What is the smallest meaningful step we can take today to test a wedge into that future?
That’s the rhythm. That’s the job.