Most business owners I speak to have already tried ChatGPT. Some use it every day. Many opened it once, asked a question, got a generic answer and decided it was not for them.
Both groups end up asking me the same thing: where do we actually start?
My answer is usually not what they expect. Start with the work, not the tool.
Why starting with tools stalls
New AI tools come out every week. A common pattern in SMEs goes like this. Someone watches a demo, the team signs up for two or three apps, and a month later nobody is using them.
The tools were not the problem. Nobody decided what job each tool was supposed to do, who would use it, or how the results would be checked. Without that, AI becomes one more login people forget about.
Here is the order I recommend instead.
Step 1: List the tasks that eat your week
Ask each person in the team to write down the tasks they repeat every week that involve reading, writing, summarizing or reformatting information. For example:
- Replying to similar customer enquiries
- Preparing quotations and follow-up emails
- Turning meeting notes into an action list
- Compiling a weekly or monthly sales report
- Translating material between English, Malay and Chinese
- Writing product descriptions or social media captions
This list is your starting map. These are the tasks where AI tends to save time first, because they are text-heavy and follow a pattern.
Step 2: Pick one task and one person
Do not roll AI out to everyone at once. Choose one task that is done often, is low risk if something goes wrong, and is easy to check.
Give it to one person for two weeks. Before they start, note roughly how long the task takes today. At the end, compare. You now have a real result from your own business, not a claim from a vendor's website.
If it works, that person becomes the one who shows the next colleague how they did it.
Step 3: Learn to give proper instructions
Most poor AI output comes from vague instructions. Compare these two requests.
The vague version:
Write a reply to this customer.
The clearer version:
You are a customer service executive at a building materials supplier. A customer is asking why their order is delayed. Write a short, polite reply in English. Apologies once, explain that the shipment from our supplier arrived late, give the new delivery date as 14 October, and offer to call them if they need it sooner. Keep it under 120 words.
The second one takes thirty seconds longer to write and usually saves several rounds of editing. It gives the AI a role, the situation, the tone, the key facts and a length limit. That habit matters more than which tool you use.
Step 4: Set simple rules before you scale
Before more people start using AI, agree on a few basics and write them on one page:
- What must never be pasted into an AI tool, such as customer IC numbers, bank details, staff salaries or confidential pricing
- Who checks AI-written content before it goes to a customer
- Which tools the company has approved, and which accounts to use
This does not need to be a long policy. It needs to be clear enough that a new staff member understands it on the first day.
Step 5: Then choose your tools
Once you know which tasks you are targeting and your team has the habit of writing clear instructions, choosing tools becomes much easier. You will know what you need. In many cases the free or entry-level version of a mainstream AI assistant is enough to begin with.
What this looks like in a small company
Take a typical 15-person trading company. The admin executive spends every Monday morning pulling numbers from different spreadsheets into a sales summary for the boss. That task goes on the list.
She tries AI on it for two weeks, using a saved instruction that tells the AI how the summary should be structured. The first week takes some adjusting. By the second week, the report is drafted in a fraction of the time and she spends her effort checking the figures instead of formatting them.
Nothing about this needed a big budget or a new system. It needed one clear task, one person and a bit of practice.
The real starting point
AI adoption in an SME is less about technology and more about habits. Pick the right tasks, start small, teach people to give clear instructions and set a few rules. The tools will follow.
At A Training Academy, this is how we run our AI productivity training for SME teams in Malaysia. Sessions are hands-on, built around the work your staff already do, and claimable under HRD Corp for registered employers.
If you would like to talk about where your team could start, get in touch with us.