AI
Where AI actually makes sense for small businesses
Strip away the hype and a handful of AI uses genuinely pay off for SMEs. Here they are, along with the ones to avoid.
7 min read By the Crescio team
Every second advert now promises that AI will transform your business. Most small-business owners we talk to have the same reaction: "Probably. But how, specifically, for me?"
Here is the honest answer. A small number of AI uses genuinely pay off for SMEs today. Several popular ones do not. Knowing the difference saves money and disappointment.
Where AI pays off
Answering routine questions. If your staff answer the same twenty questions all day (hours, prices, availability, directions, what is included), an AI assistant trained on your actual information can handle them on your website or WhatsApp, around the clock, and hand anything unusual to a person. This is the single most common win.
Reading documents so people do not have to. Invoices, delivery notes, application forms, ID documents, supplier quotations. AI can extract the fields you need and put them into your system. Hours of typing disappear.
Sorting and qualifying leads. When enquiries arrive from many channels, AI can classify them, capture the missing details and route the serious ones to the right person first.
Finding answers inside your own knowledge. Policies, manuals, past jobs, product specifications. A knowledge assistant lets staff ask a question and get the answer from your documents instead of asking the one person who knows.
Drafting, not deciding. First drafts of replies, quotations, job descriptions and reports, reviewed by a person before they go out. The person stays in charge. The blank page goes away.
Where AI usually does not pay off (yet)
Replacing judgement. AI should not decide who gets credit, who gets hired or what a complex quotation should cost. It can assist. It should not sign.
AI for its own sake. A chatbot nobody needed, a dashboard with "AI insights" nobody reads. If the problem can be solved with a normal form, a normal automation or a normal report, that is the better solution. It is cheaper, more predictable and easier to maintain.
Anything with poor source data. AI trained on messy, incomplete or out-of-date information produces confident nonsense. Fix the data first.
Customer-facing AI with no escape hatch. If a customer cannot reach a human when it matters, the AI is costing you business, not saving it.
The principle we work by
AI where it helps. Normal software where it doesn't.
Before adding AI to anything, we ask three questions:
- Is this task repetitive, high-volume and currently done by a person?
- Is the information it needs available and reasonably clean?
- Would a simpler automation solve it just as well?
If the answers are yes, yes and no, AI probably belongs. Otherwise, it is usually better spent elsewhere.
POPIA and AI
Customer information fed into AI tools is still personal information. Use providers with clear data-handling terms, minimise what you send, and be transparent with customers when they are talking to an assistant rather than a person. This is not optional in South Africa.
Start small
The best first AI project is narrow, measurable and boring: one type of document, one set of questions, one channel. Prove it, then extend it.
Crescio helps South African businesses identify where AI makes commercial sense and builds it into the systems they already use. Tell us where the repetitive work is and we will tell you whether AI, automation or a simpler fix is the answer.
