I've been using[1] Claude Code extensively since December and I actually have sort of the opposite view: That good judgement is the limiting factor and we need product managers more than ever. Several of the apps I've been working on have features that I don't plan to implement because I don't feel like thinking through how to make them make sense. I have a lot more of these examples but they're for a recipe app I haven't made public yet: The code is fairly trivial since it's very AI-based and almost every issue is blocked by deciding how the feature should actually work[2].
At this point, software engineering skill is definitely necessary for some things (algorithms, high-level design), but I'd say probably 90% of this was implemented with a prompt that just tells Claude what I want and and doesn't say how to do it.
This app is up to 1,300 commits and v1 is done. The only remaining work I'm interested in is an offline mode (hard for React-y reasons), improving content extraction (needs a lot of work that Claude would have trouble with on its own, and the brute-force method of using an LLM is too expensive), and making Android/iOS apps.
For example, the app automatically creates volume/weight conversions but people typically mix weights for things like flour with volumes for things like vanilla extract, but also sometimes you really do want every ingredient including vanilla extract to be a weight if you're making a huge batch
good judgement is the limiting factor and we need product managers more than ever
My own struggle with this sentiment -- as a product manager, no less! -- is that this implies that good judgment sits squarely in the product manager's domain and that somehow this excuses designers or engineers from exercising good judgment.
I can accept the idea that good judgment is an aspect of product management, insofar as a key function of product management is improving decision quality, but product management is an activity that doesn't have to be bound to a specific role or individual, just as writing code doesn't have to be bound to the software engineer role.
Product management as a skill-that-improves-decision-quality certainly won't go away, and I agree with you that good judgment will be more important than ever as execution gets cheaper and cheaper. But what a product manager looks like as an individual who is paid to be a product manager will either probably cease to exist soon or look at least vastly different because I just don't think there's defensible ROI in the typical day-to-day outputs of the role anymore.
Something shifted in December.
January was gone before I realized it’d been more than a month since I’d published anything. The pace of change is so rapid that even sprinting has felt a bit like running in place. Progress feels mundane. Learning new skills as quickly as new model capabilities drop feels like table stakes and hardly something worth writing about.
This isn’t existential dread or defeatism. I’m not frustrated by any of it. But it’s real and it’s visceral, and I don’t know how anyone who isn’t in the thick of it now is going to adapt to what will inevitably be a completely alien way of doing things. I don’t think I’m being hyperbolic, because I’ve seen first-hand how the patterns of compound engineering have spilled over into knowledge work, and the result is just so damn good. Going back at this point would be like intentionally unbundling the smartphone. Who really wants to carry at all times a book and a newspaper and a camera and a radio and a flashlight and god knows how many other knick-knacks and gizmos?
That’s not a rucksack I want to carry.
A lot of folks have predicted the end of my profession. Over a year ago, Claire Vo told a room full of product folks that Product Management is Dead.
I actually think this is okay. More than okay, I think the demise of product management as we know it is a tremendous accelerant for building valuable products at the core of viable businesses. In so many teams, we invest vast swaths of time into just managing the bureaucratic overhead associated with coordination. Sometimes it can take weeks of time and hours of meetings just to get everyone on the same page about what to consider building. Forging software has for so long been such a precious specialty that wielding it has traditionally occupied literal committees that in turn have defined people’s careers.
Is Claire saying that we’re likely to abandon the lossy hand-offs from customer problem to idea to testing a solution with them? I hope so. Everything will just move faster without folks like me gating context with our limited bandwidth. Maybe the people building businesses can get back to the act of actually building businesses.
But it’s probably gonna hurt for a lot of people before civilization settles into a new rhythm. We’re at the very beginning of a new industrial revolution. Our versions of the lamplighters and the ice cutters and the town criers will have to divert their energy somewhere else. Until they figure out where, a lot of people will probably lose their jobs.
It wouldn’t be unreasonable if I were one of them.
For a long time I’ve half-jokingly let people in on the open secret that I don’t have any truly marketable skills. I can’t really produce anything that anyone would use or pay for. People hire me because I have good judgment and I make good decisions and I can help other people make better decisions than they would have if I weren’t around.
That’s it. That’s literally the job. The rest—the ticket updating, the meeting scheduling, the spec writing, the project management—is all just clerical filler that’s about to get automated.
In 2025, the job offered a lot of value to teams. But something shifted in December.
As I type this, there’s been some perfect convergence of model capability and public awareness. There’s a flywheel effect at play, certainly, whereby models that are increasingly capable at tasks that previously required human effort earn more attention by virtue of being more capable. This is what every product builder aims for at the start: word-of-mouth distribution, user-generated content, crossing the chasm.
But there’s also something more fundamental happening within the zeitgeist. Not only are you likely to have seen several ads for AI during the Super Bowl (which itself is kind of insane), but you’re also equally as likely to feel the push of AI into your life in less direct ways. Your colleagues or clients are finding novel ways to be more efficient with Claude. One of your friends tried a recipe from ChatGPT that turned out great. The relatively static apps you’ve used for years are starting to sparkle and shimmer with AI features.
At first this might feel a little like the dawning of the internet or the adoption of smartphones, but it’ll become clear if it isn’t already that this is way different because the capabilities we’re unlocking are non-specific and unbounded. It’s not just that you can suddenly buy a book online or use the same device for both turn-by-turn directions and for texting your friend; it’s so much bigger than that because with some trivial investment in installing a few tools on your computer and a little bit of patience with trial and error, you can create almost any piece of software you can imagine.
In a world that runs on software, that is incredibly powerful.
Couple our civilization’s new capability with the businesses we’ve built around the creation, commercialization, and distribution of software, and suddenly new roles begin to emerge. People who were in November incredibly capable software engineers are now incredibly capable orchestrators of machines that are capable software engineers. Many of these folks were already expert strategists and shrewd business folks in their own right, but their time and attention were occupied by hand-writing code. They’ll still (for now) spend a lot of intellectual effort orchestrating the machines, but much of the organizational gatekeeping that built up around protecting their time is now effectively useless.
They can join more calls with customers. They can create prototypes for user testing. They can write their own tickets. Or write agents to write their own tickets.
If you’re building a business in 2026, where would you rather start? With a person who can simultaneously make good decisions about shaping the product while building the product? Or with a person who needs another person to write a ticket to act?
The answer to this question is obvious and was also true in 2025, but, like I’ve been saying, something shifted in December.
You’re probably wondering, okay, so what happened in December that he keeps blathering on about? And honestly, I’ve been wondering the same thing. There wasn’t any single release, a big-bang capability announcement, or blockbuster new AI product that suddenly changed things.
Nevertheless, the vibes did shift. Enough change in the way we work in tech compounded that suddenly the landscape ahead of me looked a little bit alien. The fatigue associated with keeping up escalated.
But the most prominent shift, or perhaps the thing that now stands out in much starker contrast, is the gulf between what’s transforming in my professional sphere and what’s staying the same in other segments of society. I don’t mean that in the hype cycle sense, but rather that the modern tech sector is leaving a lot of the world behind.
In tech, we have more knowledge work than ever happening in the terminal because of how damn good Claude Code is, while most people who are using AI for their jobs at all are still confined to ChatGPT (or worse, Microsoft Copilot). The most effective tools for knowledge work aren’t specialized or trade secrets, but how many public school teachers have created Claude Cowork skills to help organize lesson planning? How many small business owners are using Lovable to build and maintain their websites? How many middle managers are using NotebookLM to create their slide decks?
I’d bet a lot that the proportion is really, really small.
We can lament the lack of role clarity and mourn the death of bureaucracy in tech all we want, but the real tragedy is bigger than that: things are moving so quickly that most of society doesn’t have meaningful onramps to tools that can make their lives materially easier. Not just, “oh this is neat” easier but “oh f**k I didn’t know this was possible” easier.
So, yeah, something shifted in December. Extraordinary became mundane because everything’s changing so quickly, but it’s changing so quickly that extraordinary new things aren’t making their way into the hands of ordinary people. The profound tragedy of the situation is that the innovation isn’t isolated to specific sectors like manufacturing in the first industrial revolution. You don’t have to buy new hardware and reorganize your life like when the smartphone changed the world.
Because this change is more fundamental, more ubiquitous, the world will require more pioneers than ever before, and for those pioneers to serve as guides for the rest of us. If you’re reading this, that’s probably you.