AI just made almost everything buildable, which raises a question for every product team: if anyone can build anything, what is actually scarce now? The answer is judgement, knowing what to build, who it is for, and why it matters.

Consider how quickly the cost of building collapsed. The share of Google’s new code written by AI climbed from about a quarter in late 2024 to roughly half a year later, and Sundar Pichai now puts it at 75 per cent. When a working version of almost anything can be spun up in an afternoon, building stops being the thing that separates good teams from the rest.

So if building is no longer the hard part, then the judgement calls are, and we make them constantly, at every zoom level. What is changing is the sheer volume, because AI keeps handing us more options to weigh at every turn.

There is a deeper reason too. Something that looks like a product is not the same as something that is one, and that gap is exactly where judgement lives. As Kendra Vant says, without a reliability layer it is simply a prototype. Closing that gap, deciding what reliable enough means, what to ship and what to hold, is a judgement call. AI can produce the thing. It cannot decide whether the thing is actually ready.

This article is about those calls, the ones we make every day without naming them, how we get better at making them, and why AI, far from taking our job, may be handing the real one back.

We have to move fast. Our competitors are going to be ahead of us.

We are hearing this everywhere. And it is, in most cases, completely wrong.

In the AI era, speed has become the go-to justification for skipping the work that actually matters. Skipping discovery. Skipping validation. Skipping the uncomfortable question of whether the thing you’re building is something anyone actually needs.

I keep thinking about what it felt like when the internet arrived in Australia. Everyone was excited that they could finally build something customers could interact with directly. Off the back of that, we built a huge number of products and services. At the start of 1995, there were a couple of hundred Australian websites. Six months later, there were thousands. By the end of the decade, hundreds of thousands. There was a rush to build, a sense of possibility, and a feeling that we were suddenly closer to our customers than ever before.

That is what this moment feels like now. Around 41 percent of Australian SMEs say they are adopting AI, and about half of all Australians have used generative AI in the past year. The curve is back.

But two things are different this time.

The first is that the tools we have now give us far more control over how we work. We can build much faster than we ever could before.

The second is harder to sit with. In some cases, our customers can now build the thing they need themselves. They are not always competitors, but the gap between what we can do and what they can do is smaller than it has ever been.

Which brings me to what product people are supposed to do in this moment.

In theory, this should be our moment. If building is cheaper and faster, then the value of deciding what to build should go up. The role should shift upstream. Less time in delivery, more time shaping direction, finding opportunities, and making calls about what is worth pursuing.

That is the hopeful version.

The more honest version is that we are not really doing that.

We have always said that understanding the customer is the most important part of the job. We have always said that identifying the right problems matters more than shipping solutions. But in practice, we have often been pulled into delivery because that is what organisations reward.

AI has not changed that. If anything, it has made it more obvious.

When building gets cheaper, the temptation to build more increases. When velocity increases, output becomes even easier to measure. So instead of moving up into the harder work of judgment and trade-offs, we often move sideways into building because it is finally fast enough to feel satisfying.

It is not just a behavioural problem. It is a structural one.

Most product people are still working in environments where strategy is unclear, incentives favour output, and decision-making sits above them. In those conditions, moving faster does not help. It just means we commit to the wrong things sooner and have less time to realise it before the next decision arrives.

So the risk is not just that we build more. It is that we ship more of the wrong things, faster.

Some of those things will quietly fail. Some will create ongoing costs that no one planned for. Some will damage trust in ways that are hard to reverse. The cost of putting something into the world does not disappear just because the cost of building it goes down.

There is another cost that matters even more.

Product people have always studied behaviour, but we also shape it. Most of the time, shaping has been at the edges of people’s lives. What they buy, what they watch, how they get around.

AI is different. It is shaping how people think, how they write, how they make decisions, how they search for information. In a very short period of time, these tools have started to influence the cognitive habits of millions of people.

That shift has happened quickly, and largely without much conversation about whether we want it.

Which raises a question that is not really about AI.

It is about responsibility.

Product people are not the only ones who shape outcomes. Executives, regulators, and organisations all play a role. But we are often the last point before something goes into the world. We make the calls about what gets built and how it shows up.

So if we are not thinking about the impact of what we ship, it is not clear who is.

There are ways to be both commercially responsible and ethically grounded. But that requires trade-offs, and most organisations are not set up to make those trade-offs easily.

That tension has always been there. AI just makes it harder to ignore.

None of this is new. That is the uncomfortable part.

We have known for years that understanding the problem matters more than shipping the solution. We have known that our job is not just to observe behaviour but to shape it. We have known that speed without direction is not progress.

AI has not changed those fundamentals. It has just increased the consequences of getting them wrong.

So I keep coming back to the same questions I always have.

• Who is this product for?
• What problem does it solve?
• What benefit does the customer receive?
• What benefit does the organisation receive by putting it into the market?

Those questions have not changed.

What has changed is how quickly we can act on the answers, and how visible the consequences are when we get them wrong.

The role has the chance to become what it should have been all along.

The question is whether the systems we work in, and the choices we make inside them, will actually let that happen.

‘Everyone’s role is being disrupted. And it’s not about any role being redundant — it’s about a fundamentally new way of working.’

This is the frame we should all be holding. But it requires being honest about what that disruption actually looks like, because the honest version is both more confronting and more interesting than the LinkedIn hot takes suggest.

At Brainmates, we want to stay ahead of anything that materially shifts the product management craft. Team structure specifically matters. It touches how product managers work, what we’re valued for, and who we are at work.

It might be early to answer the question. But it’s not too early to ask it. AI is delivering efficiency gains. Lasting structural change is a different story, and it hasn’t landed yet, at least not here in Australia. What we’re seeing is plenty of experimentation in how product teams work with Engineers and Designer, but no material change to how they’re shaped.

So, we brought together a panel from four different organisations to press on it. We agreed more than we disagreed. But where we disagreed was revealing and worth sharing.

In Australia, the picture is different. Tech employment fell by around 31,000 roles in the year to May 2025, even as the broader labour market grew. Major employers have restructured. The roles most affected have been coordination-heavy, process-heavy, and support-adjacent. And while AI gets most of the credit, or the blame, for this shift, that narrative deserves some scrutiny.

AI is certainly changing how work gets done. But the scale of recent redundancies has less to do with machines replacing people and more to do with organisations finally making decisions they’d been deferring. In many cases, “AI-driven restructure” is as much an investor story as an operational one. The underlying reality is more uncomfortable: a lot of roles that existed were always hard to justify on outcomes. They were easier to justify when growth was cheap.

That’s not a recovery. That’s a reset. And for product leaders, it demands something specific: clarity on where your function actually creates value, and the commercial confidence to prove it.

A year and a half ago, I sat down for coffee with a product leader at an HR software company. They were adamant that the team was following Silicon Valley product principles to the letter: dual-track agile, continuous discovery, empowered teams. The fail fast and break stuff mantra. The full playbook. When I asked how it was going, the answer was telling: the principles made sense, but the organisation kept getting in the way. The governance structures, the funding cycles, the risk appetite, none of it was built for the model they were trying to run. They weren’t failing because of poor execution. They were failing because they were trying to transplant an operating model into soil it was never designed to grow in.

This is not an isolated story. It is a defining pattern of Australian product practice.

Australia has a productivity crisis that demands better than borrowed answers. Recently, Australia’s most senior business leaders gathered at the AFR Business Summit to sound the alarm on exactly this problem: slow growth, low productivity, declining living standards, and a loss of international competitiveness. The conversation is overdue. The Productivity Commission reports growth has averaged just 1.1% per year over the past decade, well below our historical average. AI is being positioned as the solution, with the Tech Coucil of Australia estimating Generative AI could contribute as much as $115 billion a year to Australia’s economy by 2030. But we are about to squander that opportunity by reaching for someone else’s playbook again, importing it wholesale, and wondering in five years why the gains never came. To be clear, the productivity gap is not closed by moving faster or adopting better tools. It is closed by making better decisions about what to build in the first place. To understand why, it helps to understand where the playbook came from.

The modern product operating model was forged in a specific crucible: high-growth US software companies flush with venture capital, operating in a domestic market of 340 million people, selling software at near-zero marginal cost, in a culture built around risk-taking and tolerance for failure. StartupBlink estimates the United States accounts for 46.6% of measured global startup activity. That is not just an economic statistic. It reflects an entire cultural infrastructure around entrepreneurialism that shapes even heavily regulated American organisations.

The lean startup. The minimum viable product. Growth hacking. These are not universal truths. They are solutions to problems that Silicon Valley had. Exporting them globally as gospel was always a category error. And yet here we are, still treating them as the standard against which Australian businesses measure themselves.

Australia looks nothing like California. Our economy generates most of its value through services such as healthcare, financial services, education, and government, delivered by people within systems shaped by regulation and genuine public accountability. The Tech Council of Australia estimates the tech sector contributed $167 billion to the economy in FY2021, equivalent to 8.5% of GDP. A failed product decision in healthcare affects patient safety. In financial services, it triggers APRA. In government, it costs public trust. And you don’t get to run another experiment to recover it.

The structural difference that matters most, however, is market size. Every US company operates within a domestic market of 340 million people. A failed experiment is recoverable. Australia’s market is 27 million. One significant misstep, a regulatory breach, a reputational incident, can be genuinely existential. The Optus data breach of 2022 is the clearest recent example: damaging anywhere, but in a market of 27 million people, it became defining. The mathematics demands it. And when Australian organisations import Silicon Valley’s experimentation culture without accounting for that asymmetry, they are not being brave. They are being reckless.

For too long, the response has been embarrassment. Operating within regulation, navigating complex stakeholders, and moving carefully in high-stakes environments have been framed as signs of immaturity. The constraints are not the problem. The habit of treating them as shameful is.

The organisations that are breaking the pattern haven’t found a better imported framework. They are building a deliberate, organisation-specific operating model around the real constraints they face, not the idealised constraints of a San Francisco startup.

In practice, the organisations doing the hard work of building an operating model that actually fits focus on three things.

First, honestly diagnose to determine where to focus. Most Australian organisations already know why their product efforts underperform: governance slows everything down, funding is allocated to projects rather than outcomes, and risk and compliance sit outside the product team. These are not ambition problems. They are system properties. Governance, risk, and compliance are the operating environment. Moving carefully within them is rational, not timid. No Silicon Valley framework changes that. Working honestly with the constraints does.

Second, protect discovery. The most common failure mode is solving the wrong problems, not poor delivery. The work of deciding where to invest capital gets cut when timelines tighten because it doesn’t show up on a project plan. A clear operating model makes it non-negotiable. It defines what evidence is required before investment decisions are made and gives the organisational cover to say “we don’t know enough yet.” AI has materially changed the calculus here. The research and prototyping work that once took weeks now takes days, removing the last credible argument for skipping it.

Third, borrow common principles but redesign the mechanics for your context. Guiding principles like testing assumptions early and connecting effort to measurable outcomes are universally sound. But the specific mechanics of how to do this were likely designed for different constraints. A mature operating model takes such principles and redesigns the mechanics for your environment. That might mean 6-week learning cycles instead of two, or hypothesis-driven business cases instead of detailed requirements documents. Regardless of the mechanics, the goal is the same: reducing the cost of being wrong. The path to doing that is yours to design.

Consider what made Canva successful. It wasn’t a faithful replication of the Silicon Valley playbook. It was capital efficiency born of necessity, an instinct for solving real problems with genuine discipline, and a willingness to build for global markets from day one because the domestic market was never going to be enough. They drew inspiration from Silicon Valley; they didn’t take direction from it. That distinction is everything.

With $180 billion on the table and a productivity crisis that has persisted for a decade, we cannot afford to keep making the same mistake. Product development done well by being grounded in our context, honest about our constraints, bold enough to define its own principles, is one of the most powerful levers we have because it sits at the exact intersection where customer understanding, commercial judgment, and organisational reality meet.

When a product hits the skids, the frameworks and processes meant to keep teams aligned collapse under pressure — the forces they were designed to manage become implicit, confused, or ignored. That’s when the absence of a strong internal champion in product management becomes clear.

Teams end up slipping on the boxing gloves and bouncing around in circles in the rear car park, or shipping work that skews from the desired outcome, or everyone freezes when the stakes rise and looks to the corner office. Then somebody whispers “I think it’s an execution problem”. Everyone sighs and starts scurrying around looking for a thing to fix in the process.

Product organisations already operate inside a system of tension, whether it’s named or not. The job of product leadership isn’t to eliminate that tension, because it’s an important calibration tool. It needs to be held deliberately and used smartly to ensure that ambition doesn’t become fantasy and delivery doesn’t drift.

And this is where the internal champion becomes one of the most undervalued roles in modern product work. Not the person who ‘owns the pitch.’ Not the person who keeps a programme alive because they just can’t let it go. The champion is the person who helps the team to wrestle with tension across ideas, constraints, governance, and accountability.

We recently ran some buyer research on behalf of a client, focused on platform decisions. Across interviews with senior technology leaders, one pattern held without exception: when a platform or approach progressed beyond initial interest, a credible internal advocate was there to hold the torch. Someone willing to carry responsibility for advancing it politically, operationally, and reputationally. Where advocacy was absent you could expect adoption to tank, if it appeared at all.

And it’s exactly the same dynamic that makes product work succeed or stall.

Innovation is one of those words that still appears everywhere. CEOs demand it, strategists analyse it, marketers promote it, and product managers are expected to deliver it. Yet despite all this attention, it is still uncommon to see organisations genuinely building innovation into how they operate.

Even in 2026, many organisations equate innovation with adopting new technology. Today, that technology is AI. New tools are rolled out, pilots are launched, and outputs increase. Yet, meaningful customer value and commercial impact often fail to materialise.

The issue is not a lack of technology or ideas. It is a lack of good product management.

At its core, innovation is not about tools, speed, or novelty. It is about making deliberate, valuable change. And the discipline responsible for consistently doing that is product management.

I couldn’t leave my career in the hands of an employer. So I didn’t.

As a founder of a business dedicated to building the capability of product people, I often hear that they aren’t doing any learning during the year because their organisation has no budget.

WOWSERS!

I am concerned that product people are relying on the performance of an organisation to navigate and grow their individual careers. I would never leave that up to a third party or an organisation.

You may expect product leaders and the people and culture teams to look after your personal career trajectory, but it is simply outside their remit. To be fair, it was never their responsibility to manage your whole career, only to ensure you can excel in your current role and to map out the internal pathway available to you.