Singapore’s most important AI companies are the ones nobody notices

This was first published as an Opinion piece in the Straits Times on 18 June, 2026.

Tan Kah Kee built a business empire the South-east Asian way: one viable business at a time. After his family rice trading firm collapsed in 1903, he spent his last 7,000 Straits dollars to open a pineapple canning factory at Sembawang. He knew the pineapple market from his rice trading years: demand existed, he had seen it directly, and the execution was straightforward. His cannery’s cashflow funded rubber and pineapple plantations, then biscuit, tyre, and rubber factories, and eventually a shipping line to export directly to the West.

Other local fortunes were built the same way. Eu Tong Sen’s tin mines funded a tin smelter, then a bank, and a large real estate portfolio. Aw Boon Haw’s pharmaceuticals company (the makers of Tiger Balm) paid for a newspaper network in China, Hong Kong, and Singapore, then a bank.

This South-east Asian style of entrepreneurship is grounded in context and pragmatism: start small, use available technologies, serve well-understood needs, focus on specific industries. This should be the business model for tech firms in an AI world.

A new, old kind of software business

A software developer friend in Britain recently spent an afternoon building a tool for his wife’s barristers’ chambers. It assembles electronic court submissions by bundling PDFs of documents to be submitted for a hearing, generating hyperlinked tables of contents, and formatting everything to court specifications. Previously, chambers staff spent hours preparing each submission manually. Now each bundle takes minutes to produce, with fewer errors.

Nothing about the software breaks any new technical ground. What it took was a deep understanding of how the work actually gets done by the people doing it. Just last year, building something like this would have been possible — but would have been much slower and much more expensive. The bundle-making tool would not have been economically viable to build.

But today, AI tools have made software an order of magnitude (at least) faster and cheaper to build. So small software start-ups that aim to build small software to solve real operational problems faced by small- and medium-sized non-tech enterprises have suddenly become viable companies.

That shift matters for Singapore, where nearly 80 per cent of our economy consists of traditional, non-tech industries: logistics, construction, retail, manufacturing, shipping and real estate. Most SMEs in these sectors are still struggling through digitalisation. That process is often painful and ineffective because enterprise software is either too expensive for SMEs, or forces them to change how they work to accommodate one-size-fits-all software designed for global scale.

AI tools now make it increasingly affordable to build software that does the reverse: adapting itself to existing industry workflows rather than forcing industries to adapt to software. A small team that knows freight forwarding from the inside can build the permit submission tool every forwarder files each day. A team with construction experience can build the compliance reporting tool sub-contractors run.

An AI testbed

As Deputy Prime Minister Gan Kim Yong said recently at the Future Economy Conference earlier in May, Singapore’s deep bench of SMEs is a strength and a chief reason why Singapore is one of the best places in the world to “develop, test and deploy AI solutions that solve real-world problems at scale.”

“At scale” could mean a handful of giant firms deploying monolithic platforms across global markets. But it should actually mean something else: hundreds or thousands of small software firms deploying tailored tools throughout the economy, with scale emerging from the sheer number of specialised companies solving specific operational problems.

Singapore presents different comparative advantages from Silicon Valley for AI deployment: decades-old, dense industry networks and operational expertise in logistics, finance, construction, retail, and food service; close proximity to a vast regional traditional SME ecosystem in Singapore and the wider South-east Asia region with highly specific business needs.

Germany, Austria, and Switzerland use the term “Mittelstand” to refer to a dense, successful SME sector that is distinctive for business models characterised by private ownership, long-termism, niche market leadership, and strong ties to local communities. Singapore should create a technology Mittelstand.

To do that, Singapore should focus on building highly-specialised, contextually-anchored, small start-ups that use AI tools to build products for specific business-to-business niches. This model would spread digital transformation effectively into traditional sectors, accelerate SME productivity, and create a robust technology SME population, with strong employment opportunities, that has close operational links to the traditional industries that are the majority of the Singapore economy.

A new supporting ecosystem required

But this type of AI-informed technology start-up would struggle to find investors in the current market.

For decades, technology entrepreneurship has been dominated by venture capital logic: build novel technology for enormous markets and raise venture capital to fund rapid expansion. Singapore has invested heavily in the infrastructure supporting this model, with incubators, accelerators, grants and venture funding pathways. These investments have created a blind spot. Venture capital logic treats small markets as unattractive because companies serving them cannot scale to venture-sized outcomes.

Yet, these supposedly “too small” markets are ideal for profitable software businesses of the type described above. These markets are large enough to sustain strong cashflow, yet small enough to deter competing entrants. With AI-supported code generation, tightly-scoped software that previously required months of development time and tens of thousands of dollars can increasingly be built in a fraction of the time and for a fraction of the cost, especially when solving well-understood operational problems using established approaches. But Singapore’s technology entrepreneurship ecosystem gives these firms short shrift.

SysFreight has spent 25 years building freight-management software for customs documentation, trade permits, and freight operations across South-east Asia. Yumstone’s small team has decades of F&B operational experience and provides front- and back-office software to more than 300 operators in Singapore.

Neither company appears prominently in start-up media coverage. Neither seems to have raised venture capital. These software businesses are invisible because Singapore’s technology ecosystem has been built around firms pursuing venture-scale outcomes. Yet both have survived multiple economic cycles by understanding the industries they serve from the inside.

Sophisticated investors elsewhere have already recognised the value of these small software companies. Canada’s Constellation Software became one of the world’s highest-performing listed tech companies by acquiring small software firms serving industries such as libraries, dental practices, golf courses and funeral homes. Constellation didn’t invent this category; it just recognised that such companies were systematically undervalued because they did not fit traditional venture capital expectations.

Individually, such businesses seem mundane. Collectively, they form a resilient and profitable category of software company because they are deeply embedded in real operational workflows.

What Singapore must do

Today, search funds and roll-up funds increasingly hunt for these businesses, while even some venture investors appear to be quietly shifting toward incubating them. But Singapore’s policy infrastructure has not yet adapted to this shift.

The ecosystem for smaller, cashflow-oriented software firms barely exists. What they need is patient developmental support through the early stages of the lifecycle; analogous to what venture capital incubators like Y-Combinator or Antler provide, but for founders building small, pragmatic software businesses instead of venture start-ups aiming to be the next unicorn.

The first infrastructure gap is imagination. Young founders still see only one dominant model of technology entrepreneurship: raise venture capital, target massive markets and pursue explosive growth. The idea that a small, profitable software company serving a narrow industry can itself be an ambitious and valuable tech business is still uncommon.

At the same time, SMEs often default to procuring software from large consultancies or enterprise vendors when smaller specialist providers could serve them better, faster and more cheaply. Neither side of the market can currently see the transaction that would benefit both.

The second infrastructure gap is resourcing at the earliest stages of start-up life. A freight-forwarding veteran with deep industry expertise may now have easy and inexpensive access to tools to build valuable software for the sector. But there is still no meaningful early-stage support pathway for that founder if they are not pursuing a venture-scale outcome.

Three policy shifts would make an immediate difference.

First, reform entrepreneurship grant criteria. The Enterprise Development Grant and Startup SG Tech support should be made available to companies that don’t take venture capital, with cashflow-positive growth recognised as a legitimate milestone alongside headcount growth and fund raising.

Second, recognise more types of technology founders and technology companies. Entrepreneurship support programmes should create pathways for founders whose primary credential is operational industry expertise rather than technical pedigree. And Digital Enterprise Blueprint partnership opportunities should extend beyond global software incumbents like AWS or Microsoft to include small local specialist providers.

Third, actively match-make SMEs and specialist software firms. Trade associations receiving public funding should actively broker relationships between member SMEs and specialist software builders rather than merely maintaining approved vendor lists. Productivity Solutions Grant funding should also flow more toward providers of software products that can serve a whole vertical than to one-off consultancy engagements whose benefits are not designed to replicate.

Singapore’s AI tech economy won’t be built by just a handful of giant companies winning global markets but also by thousands of small companies embedding machine-enabled and -automated workflows into the ordinary processes of the enormous, real non-tech economy: the freight forwarder’s permit system, the barristers’ bundle tool, the F&B operator’s inventory reconciliation tool. The technology to do this is new, but the logic isn’t. It’s how South-east Asia has always done it.

Tangent is an incubator for unconventional AI-informed entrepreneurship in Singapore. Find out more at tangent.org.sg.