Every SME leadership team is currently being told, from every direction, that they need an AI strategy. Vendors are pitching it. Competitors appear to be adopting it. Boards and advisors are asking about it. The pressure to "do something with AI" is real — and it's exactly the pressure that leads businesses to buy a tool first and figure out what to do with it later.
That order of operations rarely works. A tool bought without a clear workflow behind it tends to get used once, enthusiastically, and then quietly abandoned a few weeks later when it doesn't fit how the team actually works. The businesses that get genuine value from AI tend to do the opposite: they start with a specific, recurring problem in the business, and only then ask whether AI is the right way to solve it.
Start by Mapping the Workflow, Not the Market
Rather than asking "which AI tools should we adopt," a more useful starting question is: where does this team lose the most time to repetitive, low-judgment work? Common candidates in SMEs include drafting first versions of routine marketing content, summarising customer feedback or support tickets, preparing recurring internal reports, answering the same category of customer question repeatedly, and searching for information that already exists somewhere in the business but isn't easy to find.
Once a specific workflow is identified, the question of which tool — if any — makes sense becomes much easier to answer, because there's now a concrete task to evaluate it against, rather than a vague ambition to "use more AI."
Not Every Workflow Needs AI
It's worth being honest that some of what looks like an AI problem is actually a process problem. If a report takes six hours to prepare each month because the data lives in four disconnected spreadsheets, the first fix is usually consolidating the data — AI on top of a broken process just produces a faster, more confusing version of the same broken process. AI is most valuable when the underlying workflow is already reasonably well defined, and the task is genuinely repetitive enough to be worth automating or augmenting.
The question isn't "how do we use AI." It's "where is this team losing time to work a machine could reasonably help with" — and AI is only one possible answer to that.
A Simple Way to Prioritise
When a few candidate workflows have been identified, they can usually be ranked on two things: how often the task happens, and how much judgment it genuinely requires. High-frequency, low-judgment tasks — first-draft content, routine data summaries, repetitive customer questions — are typically the best starting point. High-judgment tasks, like final pricing decisions or sensitive customer situations, are usually better left with people for now, even if AI can assist in preparing the information behind the decision.
Starting small, with one or two well-chosen workflows, and proving real time savings before expanding, tends to produce far better long-term adoption than a broad initiative announced across the whole business at once.