I'm a father of three from Sydney, a Product Director and a Product Coach. I write about product management and run the Product Manager community.
Subscribe to receive digest emails (1 per month).
AI & Product
For most of the last decade, the hard part was building. You needed engineers, estimates that didn't blow out, and enough runway to reach the market before someone else did. That constraint is basically gone now, or going fast enough that the gap is closing on a timeline most people haven't internalised. Building is cheap, at least compared to what it was. What's expensive now is figuring out whether anyone will care, to whom it matters and why it matters enough to compete for their attention in a market that just got ten times more crowded. That's a product problem. And it changes what product management is actually for.
When building costs collapse, supply explodes. Every niche gets ten products. SaaS that used to need a team of twelve now takes a weekend and someone who's decent at prompting. The things that used to protect a business - the technical complexity, the time it took to ship, the first-mover window - those things are eroding fast. What replaces them is harder to copy. Distribution. Trust. Something like an audience that would follow you to the next thing you built. I've watched well-funded teams burn through eighteen months trying to out-feature a competitor who had a better newsletter, and it's a specific kind of painful to observe.
The PMs and founders navigating this well are not the ones using AI hardest. They're the ones who figured out what it cannot do: make someone care. The gap between "I shipped this in a weekend" and "people actually use this" is still enormous, and it lives entirely in distribution, positioning and product judgment. I think that's the right frame, at least. Though I'll admit I'm still working through whether I'm making sense of a structural shift or just rationalising my own situation and calling it a thesis.
Not the good ones, and I say that having thought about it enough that I'm reasonably confident it's not cope. AI is replacing the task management layer of the role: ticket writing, meeting summaries, spec drafting. What it cannot replace is the judgment to know which problem is worth solving, the instinct to kill a bad bet before engineering touches it, and the ability to hold the commercial picture and the customer picture in your head at the same time. The PMs most at risk are the ones whose value was mostly coordination and documentation. Most of them probably already know it. The ones least at risk are the ones making the actual calls.
Should product teams be using AI tools right now?
Yes, but thoughtfully rather than reflexively. Using AI to move faster on things that already matter is real leverage. Using it as a substitute for knowing what matters is just noise moving faster, which might actually be worse than slow. Before adopting any AI tool in a product workflow, it's worth asking whether it frees up time for judgment or just creates more output that still needs judgment applied to it. Most of the time it's the second one and people talk themselves into calling it the first.
How does AI change product-market fit?
It makes the supply side of every market more crowded, faster than the demand side can absorb. Finding PMF used to partly be a function of getting to market before competitors could copy you. That window is shorter now, shorter than most founding teams are pricing into their runway plans. What's become more important is something like distribution fit: whether the people who need your product can actually find it, trust it and talk about it. Technical differentiation is harder to sustain than it used to be. Audience and trust are harder to copy than any feature you can ship.
What's the difference between using AI and being AI-native?
Using AI means adding it to an existing workflow. Being AI-native means designing the workflow around what AI can do, which sounds like a subtle distinction until you see it in practice. An AI-native product team might prototype on a Tuesday afternoon, run five experiments a week and treat a weekend build as a real test rather than a proof of concept. The difference shows up in speed of learning, not just speed of output. Most teams haven't made that jump yet. I haven't fully made it either, honestly.
Does AI change what makes a good product strategy?
The fundamentals don't change. A good strategy still means making a real choice about what you will and won't do, and holding that choice under pressure. But AI makes execution cheaper, which means bad strategies get tested faster and fail faster. That's actually useful if you're paying attention. The teams that will compound advantage are the ones who can run more bets, learn quickly from the ones that fail and stay disciplined about doubling down only on the ones that show real signal. The ones getting left behind are mostly treating every failed experiment as a reason to change direction rather than sharpen it.
If you're a founder trying to work out what AI means for your product direction, or a PM trying to figure out what this shift means for your role, I'm happy to talk. No pitch. Just a thirty-minute conversation about where you are and what's worth focusing on.
My take
Most teams are using AI as a speed layer. Faster specs, better meeting notes, tickets drafted in seconds rather than minutes. All of that is real and I'm not dismissing any of it. But treating a structural shift like a productivity tool is a bit like putting new tyres on a car and calling it a different vehicle.When building costs collapse, supply explodes. Every niche gets ten products. SaaS that used to need a team of twelve now takes a weekend and someone who's decent at prompting. The things that used to protect a business - the technical complexity, the time it took to ship, the first-mover window - those things are eroding fast. What replaces them is harder to copy. Distribution. Trust. Something like an audience that would follow you to the next thing you built. I've watched well-funded teams burn through eighteen months trying to out-feature a competitor who had a better newsletter, and it's a specific kind of painful to observe.
The PMs and founders navigating this well are not the ones using AI hardest. They're the ones who figured out what it cannot do: make someone care. The gap between "I shipped this in a weekend" and "people actually use this" is still enormous, and it lives entirely in distribution, positioning and product judgment. I think that's the right frame, at least. Though I'll admit I'm still working through whether I'm making sense of a structural shift or just rationalising my own situation and calling it a thesis.
What AI means for how software is built and sold
- AI Just Removed the Waiting. Now What? - What happens when individual speed becomes universal and every niche gets ten products at once. The structural shift most teams are treating as a scheduling problem.
- AI Agents Don't Care About Your Polished UI - The era of competing on interface is ending faster than most PMs have noticed. What it means to treat your API as the actual product.
- Your Users Are Becoming Agents - When software starts navigating on your behalf, clean CTAs become obstacles more than they are differentiators. What that does to product design.
- Vibe Coding - Why cheap, fast experimentation beats quality at the hypothesis stage, and why AI makes that argument harder to dismiss.
What AI means for product managers specifically
- No Place to Hide - AI is exposing PMs who've been surviving on templates and ticket hygiene. The ones who shape bets and ask better questions are fine. The ones who don't are already being outpaced.
- Get Comfortable Being Uncomfortable - Strategy in the AI era is messier and faster. PMs who wait for certainty before moving will manage projects, not build products.
- The Future of Product Managers - Whether the PM role survives this depends entirely on what the PM is actually doing. Execution work is vulnerable. Judgment work is not.
What AI means for founders and builders
- Building Optional Income in the AI Era - A first-person account of running side experiments when building is cheap and the real constraint is getting anyone to find it. Includes the YouTube channel I'm currently testing.
Creativity, ownership and AI-generated work
- Prompted Vision - Whether AI-generated work is "yours" comes down to intent rather than process. You're choosing the vibe, the story, the framing. That's direction. That's still creation, I think.
FAQ
Will AI replace product managers?Not the good ones, and I say that having thought about it enough that I'm reasonably confident it's not cope. AI is replacing the task management layer of the role: ticket writing, meeting summaries, spec drafting. What it cannot replace is the judgment to know which problem is worth solving, the instinct to kill a bad bet before engineering touches it, and the ability to hold the commercial picture and the customer picture in your head at the same time. The PMs most at risk are the ones whose value was mostly coordination and documentation. Most of them probably already know it. The ones least at risk are the ones making the actual calls.
Should product teams be using AI tools right now?
Yes, but thoughtfully rather than reflexively. Using AI to move faster on things that already matter is real leverage. Using it as a substitute for knowing what matters is just noise moving faster, which might actually be worse than slow. Before adopting any AI tool in a product workflow, it's worth asking whether it frees up time for judgment or just creates more output that still needs judgment applied to it. Most of the time it's the second one and people talk themselves into calling it the first.
How does AI change product-market fit?
It makes the supply side of every market more crowded, faster than the demand side can absorb. Finding PMF used to partly be a function of getting to market before competitors could copy you. That window is shorter now, shorter than most founding teams are pricing into their runway plans. What's become more important is something like distribution fit: whether the people who need your product can actually find it, trust it and talk about it. Technical differentiation is harder to sustain than it used to be. Audience and trust are harder to copy than any feature you can ship.
What's the difference between using AI and being AI-native?
Using AI means adding it to an existing workflow. Being AI-native means designing the workflow around what AI can do, which sounds like a subtle distinction until you see it in practice. An AI-native product team might prototype on a Tuesday afternoon, run five experiments a week and treat a weekend build as a real test rather than a proof of concept. The difference shows up in speed of learning, not just speed of output. Most teams haven't made that jump yet. I haven't fully made it either, honestly.
Does AI change what makes a good product strategy?
The fundamentals don't change. A good strategy still means making a real choice about what you will and won't do, and holding that choice under pressure. But AI makes execution cheaper, which means bad strategies get tested faster and fail faster. That's actually useful if you're paying attention. The teams that will compound advantage are the ones who can run more bets, learn quickly from the ones that fail and stay disciplined about doubling down only on the ones that show real signal. The ones getting left behind are mostly treating every failed experiment as a reason to change direction rather than sharpen it.
If you're a founder trying to work out what AI means for your product direction, or a PM trying to figure out what this shift means for your role, I'm happy to talk. No pitch. Just a thirty-minute conversation about where you are and what's worth focusing on.
