I spent three hours last week iterating on a prototype. Moving buttons. Tweaking layouts. Adjusting copy. Shipping revisions back and forth. Then I stopped and asked myself: what am I actually trying to solve for?
I didn’t have an answer. Not a clear one, anyway.
That’s when it hit me. The problem wasn’t the prototype. The problem was I skipped the thinking that should have come before it.
I see a lot of articles right now celebrating the death of the PRD. The pitch is seductive: AI tools are so fast now, why slow down with documentation? Just build. Iterate. Ship.
But that logic confuses speed with direction. You can move fast and still go nowhere.
I use Claude heavily in my workflow. It accelerates execution in ways that still surprise me. But Claude cannot tell me what problem I am solving. It cannot tell me who I am solving it for, or why that matters, or what success looks like when I get there. That is my job. That has always been my job.
When I skip the thinking and jump straight to a prototype, I am not saving time. I am borrowing it from a debt I will repay later, usually in revision cycles, in misaligned stakeholder feedback, in a demo that looks impressive and solves nothing.
Call it a PRD. Call it a brief. Call it notes on a napkin. Whatever format works for you (it doesn’t need to be more than a couple of pages). But the thinking has to happen first. When I force myself to answer the hard questions before the build starts, everything downstream gets faster. What is the user trying to do? What are we not building? What does done actually look like?
With Claude, getting that clarity down is no longer the slow part. I can move through a draft faster than I ever could before. But fast or slow, the thinking still has to happen. Claude helps me execute. It does not replace the judgment required to know what is worth executing.
The product managers skipping requirements because they have access to AI are not moving faster. They are just generating slop with better tooling.
Is the PRD dead?
No. You just have fewer excuses not to write one.