Most AI advice aimed at small businesses is written for enterprises with a data team and a budget to match. The reality for a ten-person company is different: you don't need a moonshot. You need to get your week back.
The promise worth paying attention to is not "replace your staff." It is leverage: letting a small team produce what used to take a much larger one, by handing the dull, repetitive parts to software that is genuinely good at them now.
Start with the work nobody wants
The best first candidates for AI are not glamorous. They are the tasks your team already resents:
- Triage and routing. Sorting incoming email, enquiries or tickets and sending them to the right place with a draft reply attached.
- Summarising and extracting. Turning long calls, documents or threads into the three facts someone actually needs.
- First drafts. Proposals, listings, replies, reports. Not the final word, but a strong starting point that turns a blank page into an edit.
- Data tidying. Reconciling, reformatting and checking the things that currently live in a spreadsheet and someone's patience.
None of these replace a person. Each one gives a person back an hour, and for a small team an hour a day is significant.
A concrete example: ticket triage
One service business was spending around 90 minutes a day sorting and categorising incoming support tickets before they could be routed to the right person. The categories were consistent (billing, technical, returns, general), but the tickets were free-form text and required a human to read each one.
We wired a small classifier into their inbox: incoming tickets are read, assigned a category and a draft first response, then dropped into the right queue. A human reviews and sends. The classifier takes a few seconds per ticket.
The outcome: the 90 minutes became about 15 minutes of review work. The team did not shrink, but the two people doing the sorting now spend most of their morning on the cases that actually need human judgement.
Where it doesn't pay (yet)
It is just as important to know where to hold back. Anything with no tolerance for error and no human check (sending money, making promises to customers, irreversible actions) is the wrong place to start. So is automating a process that is broken to begin with: you will just make the mess faster.
How we approach it
We start with your week, not with the model. We look at where your team actually loses time and pick the one or two tasks where automation pays back fastest. Small, real and measurable beats ambitious and vague.
We wire the AI into the tools you already use, put guardrails around it, and keep a person in the loop where it matters. We put numbers on what it saves from day one, so you know it is working and can decide what to automate next.
If you are a small business wondering where AI could give you back time, take a look at our AI services or tell us what is eating your week. We will help you find the first win.
