What AI can, and cannot, fix in your business

Start with the task, not the technology

Most "where should we use AI" conversations start from the wrong end. They start with the technology — a model, a demo, a headline — and go looking for a problem to attach it to. That produces pilots that impress in a meeting and die within a quarter, because the task was never the bottleneck.

The better question is narrower: which task in this business is repetitive, high-volume, and currently done by someone reading unstructured information and making a judgement call? Triage of inbound support tickets. First-pass drafting of a report that follows a template. Summarising a document nobody has time to read in full. Those are the shapes of problem AI is actually good at.

Where it reliably falls over

Anything that depends on data you do not have in a usable form. AI does not fix bad or missing data; it makes the consequences of bad data harder to spot, because the output still reads fluently even when it is wrong.

Anything where the cost of a wrong answer is high and a human is not going to check every output. If an error reaches a customer, a filing, or a decision with legal weight, the review step has to be real, not a rubber stamp — and that review step has to be costed in from the start, not bolted on after the first mistake.

A smaller, more honest question

Before any AI conversation, we ask what a simpler system would do for the same problem — a template, a checklist, a rule. If that gets you 80% of the value with none of the model risk, that is usually the right answer, and it is a cheaper one to maintain. AI earns its place when the simpler version genuinely cannot cope with the variation in the input. That is a smaller set of problems than most businesses assume, and it is worth finding out which of yours are actually in it before spending on either the technology or the change management around it.