Six years as a PM taught me to write specs, argue trade-offs, and estimate work by asking engineers. This past year I built seven apps myself, and the biggest change wasn't technical skill. It was that my sense of "how big is this ask" stopped being secondhand.
Some things I expected: prototyping is faster than describing, and a working demo ends debates that documents can't.
Some things I didn't: the last 20 percent is where all the product decisions actually live. Anyone can get an AI app to a demo now. Deciding what happens when the model is wrong, what the empty state says, when to ask a human, that's the work, and it's invisible until you've had to do it yourself.
The practical effect on my day job: I scope smaller, I prototype before I spec, and I trust my estimates because some of them are mine now.