A product manager started coding a product, then he stopped, and the project accelerated.
I found this example really interesting, from a recent talk by Tamar Yehoshua at Lenny Rachitsky's event.
She explained that a PM and a designer built the first version of a new Atlassian product with AI, all the way to a working alpha that people were using internally. Only once they saw the potential did they bring in the engineering team.
But one thing happened: the PM stayed deep in the code. He then noticed the team starting to lose focus and direction while he was busy coding.
As soon as he stepped back and shifted his effort to direction and prioritization, the project started moving way faster.
Having coded the product first was still valuable: the PM understood the technical challenges much better and unblocked the team much more efficiently.
There's a lot of talk about blurring the lines between roles: PMs becoming devs, devs becoming PMs, designers shipping code. But it's hard to say where the line actually is, and I think Tamar's example draws it really well.
The PM does have an advantage coding early, in the discovery phase, when an idea needs to become concrete and hypotheses need testing. That's not what he does forever. As the project evolves, helping the team drive impact is what makes the difference.
When I see this discussed on LinkedIn, it usually sounds like a constant activity, as if the PM who codes now codes all the time. That was the main insight for me: it depends on the phase. The toolset grows, but the responsibility for direction and results stays.
The question I keep asking myself: with everything I can do with AI now, what is the most impactful work I should be doing right now for my team?