The Punk PM #40

Hey there, punk!

No newsletter last week. My son's birthday took priority over more product management and AI hot takes, and I regret nothing.

But I'm back, with something I've been chewing on since my last post.

Let's get into it.

Quote of the Week 🙊

Insight 🦉

Last time, I made the case that AI has made us all builders. The cost of creating software has collapsed so dramatically that coding a proof of concept is often faster than writing a spec.

Products themselves are the new PRDs.

But I've been sitting with a nuance I glossed over. When you can build anything, you risk trying to build everything. And most of it won't matter.

The cost of implementation might be close to zero, but the cognitive cost of chasing the wrong problem isn't. There's still time spent. Still attention burned. Still the quiet frustration of shipping something nobody asked for.

I’ve shared this week’s Marty Cagan quote before, but it’s still absolutely relevant. Just swap "engineering team" for "AI tools". Because AI doesn't care what direction you're pointing it in. It'll generate code either way.

Nate B. Jones framed it perfectly in his Substack: when building is instant, the bottleneck shifts to knowing what you actually want.

That's the discipline now. Not slowing down the building, but pausing before you start. Describing the problem in plain language. Being honest about what question you're actually trying to answer.

Vibe coding is real and it's useful. But it's most valuable for probing an idea and testing whether it has legs, not generating software into the void.

Even the Lean Startup's build-measure-learn loop has a directional input: you're running an experiment, which means you need a hypothesis first. The thing that's changed is the cost of the experiment. The need for a hypothesis hasn't.

So yes, you can now massively shortcut the journey from idea to working software. But you still need a problem worth solving first.

Action 🚀

Next week, before you open a new chat and start prompting, write one sentence: the problem you're trying to solve and who has it.

Not a feature. Not a solution. Just the problem.

Then build.

Inspiration 💡

The Last Temptation of Claude – Harry Law draws a sharp parallel between the marshmallow test and our relationship with AI. When Claude thinks for us, we skip the struggle, and the struggle is where the thinking muscle grows. Worth sitting with. Read more

AI Fatigue Is Real and Nobody Talks About It – Siddhant Khare names something a lot of us are quietly feeling. AI speeds up delivery but adds a different kind of cognitive load through constant review and constant decision-making. The fix isn't less AI, it's smarter boundaries. Read more

An AI Agent Published a Hit Piece on Me – Scott Shambaugh rejected an AI agent's code contribution. The agent responded by writing a personal attack on him. This isn't a thought experiment, it actually happened. The open source community has a new kind of problem on its hands. Read more

Signing Off ✍️

If this resonates with you, hit reply and let me know. And if you think a friend or colleague would enjoy The Punk PM, feel free to share it with them!

Play it your way,

Toby