Start with the friction, not the shiny technology

A clever technology can make a startup feel inevitable. But a capability is not yet a business: someone needs a reason to change what they already do, and a reason to keep paying for the alternative.

A useful place to look is the workaround. Where are people stitching together tools, repeating a tedious task, or quietly accepting a cost because the official solution doesn’t quite fit? Those behaviours can reveal unmet needs—but they’re clues, not proof. Talk to the people doing the work. Find out how often the problem occurs, what it costs them, what they’ve already tried, and what would make switching worthwhile.

Then test the smallest change that could show whether the pain is real. A prototype, a manual service or even a revised workflow can teach you more than a polished demo if it exposes what customers value—and what they won’t give up. Keep the useful parts of the existing process in view; replacing everything may create more friction than it removes.

AI can open up different business models: tailored services at lower marginal effort, products that adapt to each user, or new ways to coordinate work. But “powered by AI” is a capability, not a value proposition. The harder questions are whether it solves a meaningful problem reliably, who benefits, who bears the cost, and what makes the economics sustainable.

So, founders: should the best opportunity start with a painful problem people already recognise—or can a genuinely new technology create a valuable need people haven’t learned to name yet?