AI will find the pattern, but it might miss the point – How we ensure we build better digital products

Generative AI is giving product teams something they have always been short of… considered thinking time.

The hard work of synthesising business needs, customer priorities, stakeholder perspectives, product research and design insight takes serious time. AI is helping us compress that. Not replace it.

And if your organisation understands that nuanced difference, the opportunity in front of you is much bigger than a headcount conversation. It’s a conversation about allowing your greatest strategic minds, the time to contextualise your needs.

The conversation about AI for (only) efficiency is not just the wrong question; it’s incredibly reductive.

For any futurist out there, building consumer-facing digital products, these new tools will lead us to achieve the future of quality digital products: hyper-personalisation of experiences at scale.

My view is that we not only have the opportunity to achieve this, but also the standard operating practices to deliver on that promise, here is how I can see this playing out in the product strategy space.

So what does “building better digital products” actually look like?

It’s a question that is multi-facteted and probably too in-depth to cover in a single article, but here is how I see it playing out across three dimensions of product strategy.

—

1. Moving from single bets to multi-hypothesis testing

Historically, synthesising user data took so long that product teams were often forced to place all their chips on a single hypothesis. The research had to justify the time it took, which meant the output tended to be a confident recommendation rather than a range of competing possibilities worth testing simultaneously.

That constraint is gone. AI can now analyse thousands of raw customer interactions, support logs, recordings, and interaction patterns, surfacing hidden pain points at a scale that was simply not practical before. Product teams can run multiple hypotheses in parallel, learn faster, and make better-informed prioritisation calls without betting the roadmap on a single read of the room.

But the human role does not shrink here. It sharpens. Someone still has to question and validate the outputs, catch the outlier that the synthesis missed, and make the judgement call about which pattern actually matters commercially. The tool finds the signal. The product manager decides what to do with it.

2. True agility, with guardrails

Boyd’s Law of Iteration tells us that the number of iterations beats the quality of any single one. AI can drastically reduce the cost of each iteration, and that changes the product development calculus in a meaningful way.

We all know the “Spotify skateboard” concept: ship something small, learn, and iterate. The honest reality is that teams rarely follow through on this, because even skateboards take serious human effort to build. People get attached. The sunk cost kicks in. The MVP becomes the product.

With AI-built prototypes, throwing something away and starting again is far less painful. That should give us permission to get back to what an MVP was originally meant to be: a learning vehicle, not a launch vehicle.

The guardrail that matters here is discernment. Speed without validation is not iteration; it is churn, faster. Lines of code per hour is a vanity metric. Time to value, truly understood, requires human judgement at every stage to make sure you are not just building the wrong thing faster. Building the right thing, and building it right: both still require craft.

3. The hyper-personalisation payoff

This is where it gets genuinely exciting for anyone thinking about the next generation of digital products.

If the efficiency gains from AI across the development lifecycle free up human effort from production work, that effort can be redirected toward earlier, higher-leverage stages of the process. Research, synthesis, strategy, the decisions that shape everything downstream. And as teams develop a richer, more continuous understanding of how customers actually behave, the possibility of building software that adapts dynamically to individual users becomes real rather than aspirational.

We are moving away from monolithic, one-size-fits-all applications toward something far more interesting: products that meet people where they are, that evolve as usage patterns evolve, and that carry genuine intelligence at the level of the individual experience.

That is not a consequence of AI replacing product thinking. It is the consequence of product thinking, freed from the mechanical burden that previously consumed most of it.

The question worth asking is not whether AI will eliminate roles in product and design. It is whether teams will use these tools to do the same things slightly faster, or to do fundamentally better things with more context.

The product managers who thrive in this environment will not be the ones who use AI to accelerate the process. They will be the ones who use it to buy back the thinking time that good product decisions actually require. Less time transcribing. Less time on mechanical synthesis. More time on the harder questions: what does this actually mean, does it change what we build next, and are we solving the right problem in the first place?

That has always been the job. It is just easier to do it badly when the process work is eating all your time.

So yes, AI will find the pattern. And that is genuinely exciting.

Just make sure someone from your product team is still asking the right question, which is: “What problem are we solving?”.