The Punk PM #58

Hey there, punk!

Earlier this week I read that Anthropic had told its growth team to hire more product managers. When one of the biggest AI companies on earth watches its engineers start shipping at three times the pace and responds by hiring more product people, you can afford to feel bullish about where this job is heading.

The article that reported it, by Amazon engineer Ishan Gupta, drew another conclusion: that the answer is engineers who think like product managers. He's onto something real. The bottleneck has moved from writing the code to deciding what's worth building. But hand all of that to the people whose instinct is to ship, with nobody there to say "enough", and you end up with Be Here Now. Potential brilliance smothered under its own overdubs. "Think like a PM" undersells the job. The part that matters is the restraint—the taste to know which ideas earn their place, and the nerve to leave the rest on the studio floor.

Which reminded me of something Tommi Forsstrom and I kicked around on Twitter years ago: that the best product managers act like record producers. They bring taste and hard-won experience to a problem, and they're opinionated enough to back one take over the rest.

Turns out that's been the throughline all along. And those skills are getting more valuable, not less.

Let's get into it.

Quote of the Week 🙊

Insight 🦉

Rick Rubin can't play an instrument. He doesn't touch the mixing desk. Half the time he's lying on the studio floor with his eyes shut while the band plays. And somehow he's behind some of the greatest records of the last forty years. The Beastie Boys, Slayer, Johnny Cash, Adele. Hip-hop and thrash metal to torch-song heartbreak. There's no thread connecting the genres. Rubin is the thread.

So what's he actually there for? It's not to man the faders. He's there because he knows which take is the one.

I've been circling this for a couple of weeks already. Restraint, constraints, knowing what not to build. It all lands here. Producing is the job now.

For years we mistook a PM's value for the artefacts they created: the tickets, the specs, the burndown charts, the ability to run the machinery. AI is quietly eating all of it. What it can't eat is the thing Rubin does on the floor—feeling which version of the idea is going to land, and having the nerve to say so out loud. Add what the thing needs. Strip out what it doesn't. Get to the one version worth shipping and refuse to settle for the nine that were nearly right.

But here's the thing: don't read "can't play guitar" as "skipped the craft." Rubin is steeped in it. He built Def Jam out of an NYU dorm room because he'd soaked himself in New York hip-hop until he understood it better than most of the people making it. He can't play a note but he can hear exactly where a track is falling short, and he's oppinionated enough to tell you. That's the bit that matters. The producer isn't a neutral orchestrator who books the room and sorts the catering. They're a genuine co-creator. They know what great sounds like, and they'll push and prod until the musicians get there.

Same goes for us. Even freed from learning to code, I'm not the builder a good engineer is after decades in the craft. Hand me an AI design tool and I'll make something passable but I still can't do what a great designer does. What I bring is judgement. Knowing what good looks like, what a well-made product feels like, and holding the room to it. That's not a lesser skill than building. It's a different one, and it's getting rarer.

And it takes craft of its own. George Martin never wrote a Beatles song. But he was a classically trained musician, and when Paul turned up with a tune and no idea what to do with it, Martin scored the strings on "Yesterday" and built the orchestral climb in "A Day in the Life." He made four young lads from Liverpool better because he had range they didn't yet have. The producer earns their place on the sofa. They don't start there.

And the best ones don't take the credit. They push the artist into the light and stay out of it. The record stands on its own, and it's the band's name on the sleeve. It's not our job to own the features, or their impact, or the win. It's to push our teams to the point where the credit is theirs.

There's a comfortable line going round that AI will match our taste soon enough. That a model can learn what's good as well as we can. Maybe that’ll happen. But taste was never the whole story. Rubin's own word for what he sells isn't taste. It's confidence. The nerve to form an opinion and commit to it while everyone else is still hedging and waiting for the data. You can't A/B test your way to a great product. Sooner or later someone has to shut their eyes and say: that one. The PMs who matter are the ones who can. And the ones who do it well have done the reps to earn the quiet confidence that they're usually right.

That's what Rubin brings to a record. It's what the best PMs bring to their products.

Action 🚀

Next week, take something your team or your tools have built and sit in the producer's chair. Don't add a thing. Ask one question: what is this actually for? Then strip out everything that isn't serving the answer.

It'll feel like you're doing less. You're not. Deciding what the thing is and holding the line on it is the whole job now.

Inspiration 💡

Defining Taste – Mitchell Hashimoto argues that taste is the skill of making good choices without a rulebook—hard to build, easy to copy. And the copying is the tell: when everyone can make anything, the only person worth copying is the one who knew what to make in the first place. Taste isn't decoration on top of the work. It's what decides which work gets made. Read more

Craft is Untouchable – Christopher Butler makes the case that craft was never about the tools—it's the practice, the slow accumulation of judgement that teaches you what good actually feels like. AI makes it easy to skip that and settle for convenient, and the cost shows up later as quality you can't quite name the absence of. The counterweight to this week's Insight: taste buys you the producer's chair, but only craft earns it. Read more

Trusting your own judgement on 'AI' is a huge risk – Baldur Bjarnason warns that AI is the one domain where your gut is least trustworthy, because tinkering with the tools breeds confidence far faster than competence. It cuts directly against everything I've argued this week, which is exactly why it's here. Conviction and self-delusion feel identical from the inside. 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