Most writing about AI and small businesses sits at one of two extremes. In one version it replaces the whole team overnight. In the other it is a glorified autocomplete. Neither helps an owner decide whether any of it is worth their time.
The useful answer is narrower than the headlines and more practical than the sceptics allow. AI is only useful when it does something specific.
What it does reliably.
The dependable uses share a shape: work that arrives often, has a clear input and should end in a predictable output. Enquiries that follow a familiar pattern. Follow-up that needs to go out on a schedule. Documents that need reading, summarising or checking against a record. Appointments that depend on matching availability. Reports that someone currently rebuilds by hand every week.
None of that is dramatic, and it does not need to be. Configured carefully and reviewed as the business changes, it returns time to the work that needs people: conversations with customers, proposals, decisions.
Adoption is moving in this direction. Intuit's 2026 QuickBooks AI Impact Report, which surveyed more than 34,000 small and midsize business owners, found that 77% of US businesses now use AI regularly, up from 48% in July 2025, and that around 78% say it has made them more productive. These are US figures, so they describe a direction rather than a UK benchmark.
Where it stops.
AI handles repetition well and judgement badly. Reading a room, deciding on incomplete information, steering a client through a difficult moment: these belong to people. The two common mistakes mirror each other. One is expecting AI to handle what it cannot. The other is being so cautious about that limit that it never handles the work it clearly can.
A practical test: if a new member of staff could learn the task from a clear written process, AI can probably handle it inside rules the business sets. If the task depends on years of knowing particular clients, it should stay with a person. Most businesses have more of the first kind of work than they expect.
Where to start.
No technical background is needed. What is needed is a clear picture of where time goes and which parts of it need judgement. Ask Astro starts there, and shows which of Astro's systems would fit, if any.