The hot take version is that AI is either about to replace most knowledge workers or is a slightly better search engine, depending on who's writing. Neither is right, in my experience. The more accurate story is that AI has restructured what's valuable in my work — and raised the stakes on the parts that were already the hardest.

What got easier

A lot of things got easier. Research that used to take hours takes minutes. First drafts of strategy documents, job descriptions, user interview guides — I can get to a working draft faster than I ever could before. Synthesizing large amounts of customer feedback into themes that I can then interrogate? Dramatically easier. Pressure-testing an argument by asking for the strongest counterarguments? Genuinely useful in a way I didn't expect.

For the operational and administrative surface of product work — the stuff that's necessary but not where the actual value lives — AI has been a real time reclaimer. I get that time back, and ideally, I spend it on the things that actually require me.

What got harder

The bar for judgment got significantly higher. If AI can produce a competent product strategy document, a competent user story, a competent competitive analysis — then "competent" is no longer the bar. The question is whether your thinking is actually good. Whether your framing is right. Whether the document is asking the right questions, not just answering the ones that are easy to answer.

The people around you can feel the difference between thinking that was shaped by genuine insight and thinking that was shaped by a prompt. Maybe not always, maybe not immediately — but over time, in the quality of decisions that result. AI can make average thinking sound polished. It can't make fuzzy thinking clear.

"If your thinking is fuzzy, AI makes it fuzzier faster. It just produces the fuzziness at greater scale and with better grammar."

What became table stakes

Not using AI fluently is now a liability in product work. That's not a prediction — it's where we already are. The PM who isn't integrating these tools into their workflow is like the PM in 2005 who couldn't use Excel well. It's possible to work around it, but it costs you, and it signals something about your relationship to the craft.

What remains the differentiator

Taste. Judgment. The ability to walk into a room full of smart people with competing incentives and orient everyone toward the thing that actually matters. Relationships with customers deep enough that you know what they're not saying. The instinct — built from years of being wrong about things and paying attention to why — for which risks are worth taking.

None of that comes from a model. All of it still has to come from you.

The people who will be most valuable in this next period aren't the ones who've figured out the best prompts. They're the ones who've done the harder work of developing the judgment that makes the output of those prompts worth anything.