Most SME leaders we speak with are over the AI hype cycle and into a quieter, harder question: where does this actually earn its keep inside our business? After a year of running AI engagements with companies between 20 and 200 staff, a clear pattern has emerged.
The wins are rarely glamorous. They look like a sales coordinator reclaiming six hours a week, a service desk handling double the enquiries with the same headcount, or a finance team closing the books two days earlier. Boring on a slide. Transformative inside a P&L.
Three patterns that consistently work
First, AI on top of well-defined repetitive workflows — quote generation, enquiry triage, document summarisation. The work is structured, the output is checkable, the time saved is measurable.
Second, internal knowledge assistants. Mid-sized businesses accumulate institutional knowledge in chat threads, drives, and the heads of senior staff. Wrapping that in a private assistant compounds in value the longer it runs.
Third, AI-assisted content and outbound. Not full automation — assisted production. The marketer still owns the work; the model collapses the time from blank page to first draft.
Where SMEs still get burned
The pattern of failure is also predictable: ambitious cross-functional pilots with unclear owners, no measurement, and tooling chosen before the workflow is mapped. AI doesn't fix a broken process — it scales it.
Our standing advice: pick one workflow, one owner, one metric. Ship in weeks, not quarters. Then expand from a base of evidence, not a base of enthusiasm.
