Discovering the Right Problems in the Age of AI

AI is making it cheaper than ever to build software. A prototype that once took a team of engineers three months can now be assembled in days. This is genuinely transformative — and it creates an urgent strategic problem that most organisations haven’t confronted yet.

If building is cheap, the bottleneck shifts. The constraint is no longer execution — it is discovery. The organisations that will win in the AI era are not the ones that build the most. They are the ones that build the right things.

The Trap of Cheap Build

There is a seductive logic to cheap build: if it costs almost nothing to try, why not try everything? Run experiments, ship fast, see what sticks.

The problem is that this approach confuses activity with progress. Building quickly without a clear understanding of the customer problem being solved is just expensive waste delivered faster. Teams accumulate technical debt, fragment their product surface area, and lose the thread of their strategy — all at a pace that would have been impossible a few years ago.

Cheap build amplifies both the upside of good discovery and the downside of poor discovery. The stakes are higher, not lower.

What Good Discovery Looks Like

Discovery is the practice of understanding customers deeply enough to make confident bets about what is worth building. It is not market research, and it is not a phase that ends before development begins. It is an ongoing organisational capability.

Good discovery involves:

Understanding the job the customer is trying to do. Not their stated preferences or feature requests — the underlying outcome they are trying to achieve in their life or work. Clayton Christensen’s Jobs to Be Done framework remains one of the most useful lenses here.

Identifying where customers are struggling. Problems worth solving are characterised by frequency, intensity, and inadequacy of current solutions. A problem that happens daily, causes real pain, and has no good alternative is a far better target than one that is occasional, mild, or already well-served.

Testing assumptions before building. The most common reason products fail is that they were built on assumptions that were never verified. Good discovery replaces assumptions with evidence — through interviews, observation, prototypes, and small experiments — before significant engineering investment is made.

Staying close to the outcome. Discovery is not about feature validation. It is about outcome validation. Will this make the customer meaningfully better off? Will it change their behaviour? Will they pay for it, recommend it, or come back for it?

The AI Opportunity in Discovery

AI does not just change how we build. It also changes what we can discover.

AI-enabled research tools can synthesise customer feedback at scale, identify patterns across thousands of support tickets, and surface emerging themes from qualitative interviews in minutes rather than weeks. This makes continuous discovery — staying connected to customers in an ongoing way — significantly more practical for most organisations.

But the opportunity comes with a risk. AI-generated insight is only as good as the questions being asked of it. If an organisation does not have clear frameworks for what it is looking for, more data produces more noise. The discipline of discovery — knowing what you are trying to learn, who to learn it from, and how to distinguish signal from bias — is not something AI can replace. It is something AI can amplify.

What This Means for Leaders

For product and technology leaders navigating this shift, a few things matter most:

Invest in discovery capability, not just delivery capacity. Most organisations have spent years optimising delivery — building teams, improving processes, measuring velocity. Discovery capability — the ability to identify the right problems to solve — is typically underdeveloped by comparison.

Make discovery continuous, not episodic. Discovery is not a project phase. It is a rhythm. The best product organisations have structured, ongoing access to customer insight that informs decisions at every level.

Distinguish between AI as a tool for efficiency and AI as a domain of opportunity. AI can make your existing processes faster. But it can also open entirely new categories of value that did not exist before. Both matter. Both require discovery.

Set a higher bar for problem definition. As building gets cheaper, the temptation to start from a solution increases. Resist it. Spend more time — not less — on problem definition. The quality of the solution you build is bounded by the quality of the problem you have understood.

---

The organisations that will define the next decade are not those that automate the fastest. They are those that discover the most clearly — that understand their customers deeply, identify the problems worth solving, and build with purpose rather than pace.

Discovery has always mattered. In the age of AI, it is the thing that matters most.