Claude for Product Managers: How I Actually Use It

Most of my week is reading and writing: interview notes, specs, tradeoff docs, updates for five different audiences who all want a different level of detail. That's exactly the kind of work Claude is good at, and three things stand out once you've used it on real deliverables.

The writing is the part people underrate until they've compared side by side. I've handed the same PRD brief to Claude and to ChatGPT more than once. Claude's reads like a colleague wrote it; ChatGPT's reads like a good template got filled in. On a document that has to hold a consistent structure and land in a specific tone, that difference shows up on every page.

It also actually follows a brief. Give it my section structure, the tone I want, the edge cases I already know are coming — and it respects all of it. On long, constraint-heavy prompts the others tend to drift, and I'd get back a clean doc that had quietly dropped half of what I asked for. Small thing on a tweet; expensive thing on a spec that goes to engineering.

And it flags what a good PM would flag — the compliance edge case, the hole in the data model — instead of handing me something tidy but shallow. When it doesn't have enough to go on, it tends to say so rather than invent a confident answer. In regulated fintech, a confident hallucination inside a requirements doc isn't a cute mistake. It's a review cycle you pay for two weeks later.

None of this makes it the right tool for everything. ChatGPT is ahead on image and video, voice, and quick data analysis in its code interpreter. Gemini is the one to beat if your team lives inside Google Workspace, and it wins on price and raw context size. I don't think brand loyalty is the move here — I use Claude as my writing-and-thinking workbench and reach for the others where they clearly lead.

Here's what I think this does to the job. When a decent first draft of any document is basically free, writing the document stops being the skill. That's uncomfortable if writing polished docs is quietly how you've measured yourself for years — it was for me.

What's left is the part the tool can't do for you: the discovery, the judgment about what's actually worth building, the unglamorous work of getting people aligned. My time is moving away from translating and formatting toward deciding. And there's a new kind of PM showing up — one who can prototype the idea themselves and close the loop with engineering, so scoping starts from something real instead of a hopeful paragraph.

I don't buy the line that this replaces us. The tool amplifies whatever judgment you already have; it doesn't hand you any. But the floor is rising fast, and the people who'll pull ahead are the ones who let it handle the tedious middle so they can spend their few good hours on the things that actually move the product.