Most small business owners have the same question about AI: where do we even start? There are new tools every week, and plenty of advice that assumes you have a technology team, a budget for experiments, and time to spare. Most small and medium-sized businesses have none of those.
The good news is that a useful first step is usually much smaller — and much more ordinary — than the headlines suggest.
Start with a task, not a tool
The most common mistake is choosing a tool first and then looking for something to do with it. That tends to produce a subscription nobody uses.
Instead, start with the work. Ask: what does my team do every week that is repetitive, predictable, and slightly annoying? Good candidates usually sound unglamorous:
- Sorting incoming emails or web inquiries and sending them to the right person
- Re-typing details from a form into another system
- Chasing customers or clients for missing documents
- Answering the same internal questions again and again ("Where's the form for…?", "What's our process when…?")
- Writing first drafts of routine messages and summaries
None of these are exciting. That's the point. They're frequent enough that improving them matters, and simple enough that you can tell quickly whether the improvement is real.
Signs of a good first task
When you have a few candidates, check each one against four questions.
1. Does it happen often? A task that comes up daily or weekly gives you plenty of chances to see whether a new approach works. A task that happens twice a year doesn't.
2. Does it follow recognizable steps? If you could write the steps on an index card, it's a good fit. If every case needs a judgment call from your most experienced person, it isn't — at least not yet.
3. Is it low-risk if something goes wrong? For a first project, choose work where a mistake is easy to catch and cheap to fix. A draft reply that a person reviews before sending is low-risk. An automated refund is not.
4. Is it easy to check? You should be able to look at the result and say "yes, that's right" or "no, fix this" without a long investigation.
A task that scores well on all four is a strong starting point.
Three questions to ask your team
The people who do the work every day know where the time goes. Before you decide anything, ask them:
- Which part of your week feels like copying, chasing, or searching?
- Which questions do you answer over and over?
- If one routine task disappeared tomorrow, which would you pick?
You'll often hear the same two or three answers from different people. That overlap is a strong signal.
Asking also matters for a second reason: people are far more likely to use a new tool that solves a problem they named.
What to avoid first
Some areas are worth leaving until you have experience and clear controls in place:
- Decisions that affect customers directly, such as approvals, pricing exceptions, or refunds. Keep a person in charge of those.
- Regulated or highly sensitive information, such as patient records or client financial data, unless the tools and agreements in place are suited to it. Check with your compliance advisor first.
- Big-bang projects that try to change many processes at once. If something goes wrong, it's hard to tell what caused it.
None of these are off-limits forever. They just aren't the best place to learn.
A simple next step
Pick one task. Write down how it works today, who does it, and roughly how often it happens. Then describe what "better" would look like in plain words — for example, "inquiries reach the right person the same day, with a draft reply ready to review."
That short description is enough to have a productive conversation about whether AI, a simple automation, or a different fix entirely is the right answer. Sometimes the honest answer is that a better checklist would do the job. That's a good outcome too.
If you'd like a second opinion, that's exactly what a discovery call is for. Bring the task; you don't need a plan.
