Direction · August 2026 · 4 min read
Finding the first worthwhile use case
Most organisations do not have an AI problem. They have a queue of operational problems and a new tool of uncertain relevance. The first use case decides whether the tool earns trust or burns it.
The question arrives in every boardroom eventually, usually wearing urgency it has not earned: where should we use AI? It is the wrong question, and organisations that answer it directly usually buy an expensive lesson. The right question is older and duller: what is getting in the way of the operation, and what would it be worth to fix?
The difference matters because first projects are not really about their own return. They are about calibration. A first use case teaches the organisation what this technology is, what it needs and whether the people proposing it can be trusted. Choose well and you earn the licence to do harder things. Choose badly and every future proposal pays the tax.
What a worthwhile first use case looks like
The strong candidates share four properties. The problem is real: named people lose named hours to it, or a named decision is made late or blind because of it. The value is legible: you can say what improves and how you would know. The route is plausible: the data exists, can be reached and can be trusted, and the workflow can absorb the change. And the blast radius is contained: if the first version disappoints, it embarrasses a team, not the operation.
Notice what is missing from that list: novelty. The best first use case is often the least impressive one in the workshop. A tool that drafts the monthly compliance narrative from data the team already collects will never make a conference keynote. It will quietly pay for itself, and it will make the second project easier to approve.
The queue is the strategy
A single use case is a bet. A prioritised queue is a strategy. The discipline that finds the first project, examining problems, weighing evidence, pricing feasibility and adoption risk, is the same discipline that should keep ranking the rest. Organisations that institutionalise that queue stop asking where to use AI. They ask what the operation needs next, and whether AI is part of the answer. Sometimes it is not, and saying so is the cheapest credibility an adviser can buy.
Start with what is getting in the way. The technology will find its place, or it will honestly fail to, and both outcomes are worth knowing before the budget is spent.
Written by Piers Corfield, Chief Executive Officer, Ballista.
Talking beats reading.
If you are weighing competing use cases, or being asked to find one, tell us what is actually getting in the way of the operation. That is where the answer lives.
Your message goes to the people who would do the work.