Most small businesses in Southeast Asia already run on WhatsApp. Orders arrive there, suppliers send invoices there, staff send photos of receipts there. So when owners hear “AI”, the first idea is usually a chatbot that answers customers faster.
After building AI systems for food, professional services and property businesses in Indonesia and Singapore, I think that idea points at the wrong problem. Answering is rarely what eats a team’s day. What eats it is everything that has to happen after the answer: the order that has to be written down, the payment that has to be matched, the delivery that has to be booked, the receipt that has to end up in the books.
Here is what we have learned about making AI useful for an SME, rather than impressive in a demo.
Start with one workflow, not “AI for the business”
The projects that work start narrow: one recurring job that someone does every day by hand. At Rosalie Cheese, a food business selling through WhatsApp, that job was order admin — reading chats, confirming orders, arranging payment and delivery. It took the team more than four hours a day. Once an agent handled that flow, with the team reviewing instead of typing, it took about 20 minutes.
The narrow start matters because it gives you a number you can check. “We use AI now” is not a result. “Order admin went from four hours to 20 minutes” is.
The bottleneck is the handoff, not the answer
At a professional-services firm we work with, reconciliation took a team of four more than 20 hours a week. None of that time was spent on hard accounting judgement. It was spent moving information: from a bank statement to a spreadsheet, from a receipt to a ledger, from one person to the next.
When those handoffs run on their own and only the real exceptions reach a person, the same work takes under 30 minutes a week. The accountants did not get faster at accounting. They stopped being the copy-paste layer between systems.
Also Read: Gen Z is bringing its own AI to work. Singapore’s SMEs need rules, not bans
Numbers should come from code, not from a model
A language model is very good at reading a messy photo of a receipt or a voice note from a supplier. It should not be the thing that decides what a total is. In the systems we build, the AI reads and interprets, but figures and actions follow fixed, checkable logic. If the books say a number, there is a rule you can point to that produced it.
This is the difference between a tool a finance lead will trust and one they will quietly double-check forever — which means no time saved at all.
Anything involving money or a customer waits for a human
The fastest way to lose an owner’s trust is an AI that sends the wrong thing to a customer or pays the wrong invoice. So consequential actions wait for approval. In practice that approval is one tap in the same WhatsApp thread, and it keeps the owner in control while removing the typing around the decision.
Go live in weeks, inside the tools the team already uses
SMEs do not have time for a six-month transformation programme, and they do not want another dashboard to log in to. A first workflow should be live in about a week, running in WhatsApp, the spreadsheet and the accounting software the team already has. If the new system needs its own training course, it will not be used.
Also Read: Singapore SMEs buy back time with AI, but messy data keeps pilots stuck
What to ask before you buy
If you are an SME owner looking at AI, three questions cut through most of the noise:
- Which single, recurring job will this take off my team, and how many hours does that job take today?
- When the AI is unsure or wrong, what happens, and who approves anything involving money or customers?
- Does it work inside the tools we already use, or does it add a new one?
The answers will tell you more than any demo. The goal is not to have AI. The goal is to get your team’s evenings back.
—
Editor’s note: e27 aims to foster thought leadership by publishing views from the community. You can also share your perspective by submitting an article, video, podcast, or infographic.
The views expressed in this article are those of the author and do not necessarily reflect the official policy or position of e27.
Join us on WhatsApp, Instagram, Facebook, X, and LinkedIn to stay connected.
The post Your SME doesn’t need a chatbot. It needs the work after the chat done appeared first on e27.