The Punk PM #43

Hey there, punk!

When everything's in flux, it's worth focusing on what stays the same.

This week, I've been thinking about what's still true for product teams in the AI era. Scratch a bit at the hype and you'll find the job hasn't really changed, and that's actually a reassuring thing to hold onto.

Let's get into it.

Quote of the Week 🙊

Insight 🦉

Marty Cagan says a product team has two jobs: discover the product to be built, then deliver it. AI is disrupting both. But the jobs still need doing.

There's a trap buried in all the excitement about AI prototyping tools. High fidelity creates a bias. When something looks finished, we treat it like it is. We start optimising the thing instead of questioning whether it's the right thing. A polished prototype built on untested assumptions isn't a discovered solution; it's a well-dressed hypothesis. And confusing the two can be expensive.

That's not an argument against building fast. It's an argument for working at the right fidelity for the problem. Sometimes that's a vibe-coded MVP to stress-test a risky assumption. Sometimes it's hand-drawn wireframes on a whiteboard. The craft is knowing the difference. And no AI tool can make that call for you.

There's also a sleight of hand in a lot of the current narrative around AI and software development. We're being told engineers are becoming obsolete, largely by the companies selling the tools that would replace them. It's worth keeping that in mind.

Amazon's recent AWS incident is instructive. Autonomous agents pushing to production caused a big enough issue that they've added a human sign-off gate back in. Which quietly undermines a large chunk of the productivity gains they'd been banking on. Turns out letting the robots run free has consequences.

Here's the thing: you've never needed more than a computer and a text editor to build something real. Getting to market properly, safely, in a way that creates real value still costs what it always did.

As a product manager, the tools available to you have expanded dramatically. But the job hasn't. You still need to make sure you're building something people actually care about, in a way that makes sense for your business. AI can help with that. It doesn't have to. Figuring out where it genuinely adds value, and where it just adds risk and noise, is part of the work now too.

Action 🚀

Next week, before you build anything, ask yourself: what's the riskiest assumption baked into this idea, and what's the cheapest way to test it?

If your answer involves writing code, you're probably working at the wrong fidelity.

Inspiration 💡

You Could Make One For Your Mom, Though – Shane Mac makes a compelling case for the personal app renaissance. With AI and open networks lowering the barrier to near-zero, we might be heading back to an internet of millions of small, connected, community-owned apps rather than a handful of platforms that own everything, including your data. Read more

Thoughts About Interfaces Acting Like People – Laura Michet raises an uncomfortable question about AI assistants and delivery robots that are designed to feel human. Where's the line between helpful and deceptive? She thinks we should be a lot more clear-eyed (and a lot less polite) about what these things actually are. Read more

What Design Leaders Must Unlearn to Lead in an AI-First World – Arin Bhowmick argues that the skills that got design leaders here won't get them where they need to go next. In an AI-first world, the job isn't making things look good. It's building products that feel trustworthy. That requires a different kind of thinking entirely. 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