What "digitalise" tends to mean
The word gets used for three quite different things, and a company can be far along on one while having barely started on another.
The first is putting a process into software. Accounting moved into Xero or QuickBooks, payroll into an HR system, sales into a point-of-sale terminal. This is what most grant catalogues mean by a digital solution, and most SMEs in Singapore have done some of it.
The second is getting the company's data out of people's heads and phones and into a place the company owns: the stock level in one person's spreadsheet, the approvals in a WhatsApp group, the customer history in the sales lead's memory. This kind of digitalisation is quieter and, in the SMEs we meet, it is where the real exposure sits.
The third is changing how decisions get made. Having last week's numbers on Monday, while they can still change what happens this week, and having them without someone spending Friday afternoon assembling them. This is the one owners usually want, and it depends almost entirely on the second.
Which of the three were you picturing? It is worth saying out loud, because each asks for something different: the first is mostly a purchase, the second a project with an owner, and the third a change of routine that takes months to settle.
Why the current way makes sense
Spreadsheets and WhatsApp are fast, free, and everyone already knows how to use them. They grew with the business, and the person who runs the spreadsheet is usually good at it. Nobody chose this system; it accumulated, one reasonable decision at a time, while the team was busy with customers. That is how most working companies look from the inside, and a plan that starts by calling it a mess tends to lose the people whose cooperation it needs.
There is also a real cost to changing it. A new tool means a few weeks when things are slower, a person who has to learn it while doing their old job, and the risk that it joins the previous tool nobody uses. Owners who hesitate are weighing that honestly. What tends to help is picking the one change that would pay back that disruption soonest, and in our experience it is smaller than the word digitalisation suggests.
What the numbers say, and what they leave out
IMDA's Singapore Digital Economy Report 2025 reported that 95.1% of SMEs had adopted at least one of six digital areas in 2024, and that AI adoption among SMEs rose from 4.2% in 2023 to 14.5% in 2024. More than 26,000 SMEs had been reached by programmes under the Digital Enterprise Blueprint.
Those figures say something useful: having at least one tool is close to universal, so for most companies digitalisation now means the second and third tool, and the harder question of whether the tools talk to each other. They are quieter about depth. A company with an accounting package and a website counts in the 95.1%, and so does one whose stock, orders and approvals all run through connected systems. The gap between those two is where most of the work is.
The AI figure is worth reading in degrees as well. A tripling from a small base means roughly one SME in seven had adopted AI in some form by 2024, and the report found adopters using it across an average of three business functions. A survey covered by CNA in October 2026 found that where firms use AI in HR, a human signs off 85% of the time, and that larger organisations were more likely to be using it than SMEs. That looks to us like the sensible middle of the adoption curve: AI as an assistant whose output someone checks.
Pre-approved solutions, or something built for you
IMDA's SMEs Go Digital programme offers a free Digital Health Check, an Industry Digital Plan for most sectors, and a catalogue of pre-approved solutions through its CTO-as-a-Service platform. For a commodity function, that catalogue is a good place to start. Accounting, payroll, HR, point of sale and e-commerce are solved problems, and a pre-approved tool is usually cheaper, better supported and simpler to fund than anything built for you.
Custom work earns its place in two situations. The first is a process close to how you compete: the way you quote, schedule, or handle a customer's exception is sometimes the reason customers choose you, and a generic tool flattens it. The second is glue: two tools that each do their job and refuse to share, with a person copying between them. Glue is often the highest-return project in the company and the least glamorous. Many of the first builds we do for SMEs are a single intake form, a shared table and a daily email, which is to say they are small.
Both answers are reasonable. The pattern we see most often is a large system bought to solve a small glue problem, and then half-used because the glue problem was the only part anyone needed.
Where AI fits
Our view is that AI is a tool the way a spreadsheet is a tool, and the useful question is the same one you would ask of a spreadsheet: which task, done by whom, how often. For SMEs the answers tend to cluster around drafting replies to routine customer messages, summarising long documents and chat threads, pulling fields out of invoices, delivery orders and forms, and answering questions over the company's own records, which only works once those records are somewhere an AI can read.
That last condition is why AI and the second kind of digitalisation are the same project more often than people expect. Our AI transformation for SMEs page goes into the data side; the short version is that the tools are ready and most SME data needs some tidying before it can be used safely.
Then there is the question of what is already happening. In many companies we talk to, staff are using a chatbot for work before the company has a view on it. That is a reasonable thing for staff to do and a reasonable thing for an owner to feel uneasy about. A one-page rule covers most of it: which kinds of data may go into which tools, which tools the company has a proper agreement with, and who checks output before it reaches a customer. The PDPA applies to personal data wherever it goes, so customer and staff details are the first category to look at.
Who owns it afterwards
The quiet failure in SME digitalisation is the tool that was bought, set up, demonstrated, and then used by one person, or by nobody. In our experience three things decide whether that happens: whether someone inside the team owns the system by name before it is chosen, whether the tool starts with one process and adds modules later, and whether anyone planned the fortnight after go-live, when the old spreadsheet is still open in another tab and the new tool is slightly slower because it is new.
We have underestimated that fortnight ourselves more than once, and we now treat it as part of the project. Our case study on giving every service one owner is about the same idea at a larger scale.
Paying for it
Since 30 September 2026, Enterprise Singapore's EDGE Grant has replaced the Productivity Solutions Grant, the Enterprise Development Grant and Market Readiness Assistance. A company can receive up to S$100,000 in grant support a year across all activities, and within that cap up to S$30,000 on single-function digital solutions, integrated enterprise systems and selected automation. On most activities SMEs receive support of up to 50% of qualifying costs. Applications go through the Business Grants Portal and must be submitted before you pay a vendor or start work; the grant is paid on claim after the project is done. We have written up the scheme separately on our EDGE grant page, with the EnterpriseSG sources.
Separately, CNA reported in October 2026 that more than S$45 million had been set aside to help Singapore manufacturers adopt AI, with ready-to-deploy tools for areas such as production scheduling and predictive maintenance. If you are in manufacturing, that is worth a look alongside EDGE.
A scene, to make it concrete
Picture a 25-person food supplier in Woodlands. Orders arrive by WhatsApp from about sixty restaurants, a staff member keys them into a spreadsheet each morning, stock is counted on paper, and the accountant gets a printout monthly. The owner has heard "digitalise" from the bank, from a trade association, and from a nephew.
The first step that fits is small: an order form the restaurants fill in from a link, feeding a shared table the warehouse can see. Three months later, a stock count on a phone replaces the paper, and a daily email tells the owner what left the building. Only then does AI come in, summarising the customer messages that still arrive by chat and drafting replies for the staff member to check. None of it is dramatic, and by the end the company owns its data for the first time. The supplier is hypothetical; the sequence is one we would propose to most companies of that shape.
Questions to take into your next management meeting
- Which spreadsheet would hurt most if it vanished tonight, and who besides its owner can read it?
- Which handoff goes through someone's phone, and what happens when that phone is on leave?
- What report does someone rebuild by hand each month, and who actually reads it?
- Who in the team already uses AI for work, and for what?
- If one system had a named owner by Friday, which would it be?
If it would help to think these through with someone outside the company, a short conversation is usually enough to say where the first step sits, and whether it is a pre-approved tool, something small built for you, or simply a decision about ownership.