What Can AI Take Over in a Business With Fewer Than 20 People?
In a business with fewer than 20 people, AI can take over most of the work that repeats. It can sort routine email and draft the answers. It can write first versions of posts and newsletters from your own photos and notes. It can collect receipts, prepare a quote from the last ten you wrote, keep the customer list tidy and remind you every morning of what is still open. The conversation with the customer, the price you decide and the craft itself stay with you.
I know this from my own desk. The Change Republic is a small company. There is no assistant down the corridor, no marketing team and no IT department. What I have is a team of AI agents that I built myself, and a clear line between what they do and what I do.
This article shows where that line runs, so you can find it in your own business.
The Work That Repeats
Every business has two layers of work: one that repeats and one that depends on you.
The repeating layer is larger than it feels, because each task arrives with a real name and a real deadline attached, so it looks new. Think of the supplier email that needs the same polite answer as last month, the quote that is almost identical to the previous one, the receipt that has to be found, renamed and filed, and the newsletter that gets postponed again because there is never a free morning.
Here is what my own agents do in that layer, as an example you can check against your week.
One agent collects my business receipts. It finds each receipt in my email or on the supplier's website, names the file and puts it in the right folder for my bookkeeping.
Another sends me an email at six every weekday morning. It lists which proposals are still open and who is waiting for an answer from me. It is a plain list, and it never forgets.
A third helps with meetings. It prepares me beforehand with what was agreed last time, and it turns my notes into a summary afterwards. A fourth takes a video I recorded, cuts a short clip for Instagram and drafts the caption. I look at the clip and the text, and only then does it go out.
Each of these agents started as a plain description, in ordinary sentences, of how I do the task myself. I am no programmer.
A useful exercise is to write down one real week. Every task, however small. Then mark each one that would look the same if you filmed yourself doing it three weeks in a row. Those marked tasks are your repeating layer, and it is usually bigger than expected.
What This Looks Like in a Small Business
The examples above come from a coaching and speaking business. The same pattern fits a workshop, a practice, a studio or a shop that works by appointment.
Email is the usual starting point. An assistant connected to your inbox can sort what came in overnight, draft replies to the routine messages in your tone, and leave the ones that need your judgment at the top. You read, correct and send.
Marketing from your own material comes next. Many small businesses sit on years of good photos and have no time to use them. An assistant can look through the archive every Monday and suggest three posts, with text. You choose, change a sentence and approve. Nothing is posted by itself.
The customer list is the third area. Contacts that live in an old database, in a spreadsheet or on paper can be moved into a modern system and sorted, so that a newsletter about one topic reaches the people who care about that topic.
Then there is being found. When someone asks an AI assistant for a recommendation in your trade and town, the answer is built from what the internet says about you: your page titles, your descriptions, your Google profile, the directories that list you. An assistant can write and maintain these basics much faster than doing it by hand.
One practical detail matters more than it seems. Some owners type, and many would rather talk. A setup can be built around dictation, so that you speak a rough instruction while walking through the workshop and find the draft waiting when you sit down. The tool should fit the way you already work.
What Stays With the Owner
Some things you just do not hand to AI.
The first conversation with a new customer. The moment you look at a piece of work and decide it is not good enough to leave the building. The price for an unusual job. The phone call when something went wrong. Hiring. These depend on who you are and what you notice, and they are the reason customers come to you and stay.
Approval also stays with you. In my business nothing goes to another person without my yes. My agents prepare and I decide. That rule costs me a few minutes a day, and it is the reason I trust the rest.
And the direction stays with you. An assistant will happily produce three posts a week for a product line you should have dropped. Deciding what the business does next is owner work, and the hours AI gives back are best spent exactly there: with customers, on the work that brings in revenue, and on the questions that never fit into a full day.
This is also why the decision to bring AI into a small business has to be the owner's own. In the large companies where I speak, I often see AI tools arrive from above while the people who should use them never feel the benefit in their own day. A small business has the opposite chance. The person who decides is the person who feels the difference on Monday morning.
If you want to try this without any setup at all, pick the one task you postponed this week. Describe to an AI assistant, in plain sentences, how you would do it yourself. Read what comes back. That small test will tell you a lot about what your own repeating layer could look like.
AI setup and coaching for owners of small businesses is available at The Change Republic, in Zürich and across Switzerland. Find out more at www.thechangerepublic.com/ai-setup-small-business
Tünde Lukacs is an executive coach, keynote speaker on AI and leadership, and founder of The Change Republic. She sets up AI for owner-led small businesses across Switzerland and coaches their owners to run it themselves.