Most of the AI conversation in business right now is unhelpful. It's either utopian — the agent that runs the company by itself — or alarmed — the agent that's about to take everyone's job. Neither version describes anything that's actually happening on a Tuesday morning in a real store.
The honest version is more boring and more useful. AI in operational businesses is good at three specific things, and once you focus on those, the rest of the noise becomes much easier to ignore.
What it's actually useful for
Summarising things nobody has time to read. Daily store reports, weekly performance, audit findings, customer reviews. A short, well-structured summary of "here's what changed this week and what you should look at" saves a manager an hour and surfaces things they would have missed.
Spotting patterns that are easy to miss. A specific store's waste numbers drifting over four weeks. A particular shift consistently underperforming. A supplier that's gradually become more expensive. People can find these things — but only if they're looking. AI is patient about looking.
Drafting the boring written work. SOP updates, response to a customer complaint, the weekly note to franchisees, the monthly board update. Drafts that a human cleans up are a lot faster than drafts a human starts from scratch.
That's most of it. Three categories. Each one quietly useful. None of them dramatic.
What it's not useful for
AI is not a substitute for the things managers actually do — the conversation with the team member who's struggling, the read on whether a manager has the right judgement, the call on whether to open a new site, the relationship with a key supplier. None of that is going to be done by a model.
It's also not a substitute for the operational data layer underneath. An AI summary is only as good as the data it's summarising. If the underlying systems are inconsistent, the AI just produces confidently wrong summaries faster.
The practical test
A useful AI application in an operational business should pass three tests:
- Does it save someone real time, today? Not in a future quarter. This week.
- Does it surface a decision that someone has to make? AI is at its best when it makes a human decision easier, not when it tries to make the decision itself.
- Could you turn it off without breaking the business? If the answer is no, you've built a dependency, not an assistant.
Almost all the AI work that's worth doing in franchise, retail and hospitality businesses passes those tests. The rest is mostly marketing.
The quieter promise
The genuine promise of AI in operational businesses isn't that it replaces managers. It's that it lets a small head office team behave like a much larger one — quietly handling the repetitive work, surfacing the things worth surfacing, and giving the human team more time for the conversations that actually move the business.
That's the version of AI we're interested in. It's less exciting than the hype. It's more useful in real life.