Real value vs. hype: a field guide for the AI era

Something changed in the last few years, and most of us are still catching up. Building software got cheap. Really cheap. The same tools that sit on your laptop sit on everyone else's, which means a polished demo is no longer how you stand out. It is how you blend in.

So what did not get cheap? Knowing one specific customer so well you can predict their week. Owning a channel that puts you in front of them. Proving, in numbers a CFO believes, that you moved the outcome. None of that got easier. Value migrated from can you build it to whose problem is this, and what is it costing them right now.

That argument is the core of a field guide our co-founder Andrija Ilić wrote for a guest lecture at the InFiNet Code bootcamp, where a room of developers spent an afternoon testing their own ideas against it. This note covers the three tools that did the most work in that room. The full guide, with worksheets and a worked example, is free to download below.

The thirty-second test

Describe your product out loud, without naming a single piece of technology. No "AI", no model names, no "powered by". Just what changes for a human being. Thirty seconds.

Watch what happens. The capability-first idea collapses in real time, because there is nothing left once the technology words are removed. The pain-first idea barely notices the constraint. It is a brutal little test, and it costs nothing to run.

Value is a number, or it is not value

Vague value is indistinguishable from hype. The discipline that fixes it is one equation:

Value per customer per year = the cost of the problem today, minus the cost of your solution.

Cost of the problem: hours per week times hourly cost times fifty-two, plus the annual cost of errors, fines, rework, and lost revenue. Cost of your solution: what the customer pays plus their adoption effort. A healthy business captures 10 to 25 percent of the value it creates and leaves the customer the rest as upside.

The guide walks a worked example: a public-sector web team losing about twelve hours a month to accessibility compliance, plus roughly €10,000 a year in audit findings and rework, is a €21,500 problem per organisation per year. Price a solution at €400 a month and you have captured 22 percent of the value, and the customer keeps a multiple of what they spend.

Not all value sells equally

Three kinds of value, ranked by ease of sale: saved time, then saved money, then reduced risk. Time is the hardest sell because nobody budgets for it. Risk is the quiet champion, because risk carries a deadline, and a deadline is a salesperson that never sleeps. It is the entire reason the worked example is built on accessibility law rather than convenience: the law supplies the urgency.

The map that catches hype early

Two axes, value up and effort right, four quadrants: quick win, strategic bet, filler, hype trap. The mistake almost everyone makes is scoring only the effort to build, which is the one component that got cheap. The honest formula is build plus sell plus support plus compliance. When building is cheap for you, it is cheap for every competitor, the market floods, and your sell effort inflates. That is how a flawless weekend prototype ends up stranded in the hype trap.

The useful turn is that the map is for steering, not just judging: niching moves the dot up and left without changing the underlying technology, because value rises when the product fits exactly and effort falls when the buyer recognises themselves.

If you want the interactive version, the workshop we ran at the bootcamp lives at aivalue.quince-ai.com, with the same tools as a guided worksheet you can fill in for your own idea.