This week, two numbers came out of the same region within days of each other, and almost nobody put them side by side.

On September 3, Singapore’s central bank committed SG$220 million over three years to fintech and AI infrastructure under a fourth-generation innovation scheme. Around the same time, a Tracxn landscape report showed Southeast Asia’s native AI startups had raised US$4.1 billion through July, more than double all of 2025. On paper, that reads like a boom.

Except one company, Kling AI, absorbed 68 per cent of that total in a single Series D. Strip that round out and the region raised roughly US$1.3 billion, spread across just 23 disclosed deals, down from 41 the year before. And separately, no dedicated Southeast Asia private equity fund closed at all in the first half of 2026 — the worst showing on record — while three pan-Asian mega-funds raised a combined US$39.2 billion that has no obligation to land in this region specifically.

The tool gap closed, the timing gap didn’t

Here is the part of the AI story that gets the applause: building software no longer requires a computer science degree. By one recent estimate, 84 per cent of people using AI coding tools in 2026 have no engineering background. Lovable reportedly hit US$400 million in annualised revenue this year, four times what it was eight months earlier. A non-technical founder took a Lovable-built tool to over US$800,000 in ARR in nine months.

I am one of these people. I scored 147 on the PSLE and was streamed into Normal Technical — not the track anyone expected a founder to come from. I built my first product during downtime on ambulance shifts, using AI as my technical co-architect. So I understand, from the inside, why “anyone can build now” feels like the headline.

Also Read: Singapore’s AI dividend will depend on what happens after the pilot phase

But building access was never the real gate. Information timing is, and that gate has not moved.

Institutional investors have spent years buying lead time most consumers don’t know exists. Search data has been shown to run 4 to 10 weeks ahead of reported revenue for consumer-facing businesses; card transaction data offers a 2 to 4 week head start on earnings surprises. None of that is illegal or even secret — it’s a mature industry with vendors and case studies. It’s just priced for institutions, not for the person deciding whether to buy the stock, the shift, or the property everyone will be talking about in a month.

Consumers can feel it, even if they can’t name it

Consumer AI usage keeps climbing — Prophet’s 2026 research puts adoption at 73 per cent, up from 45 per cent in early 2024. But belief that AI will be trustworthy enough to lean on for real decisions has dropped roughly 30 per cent in the same window. That is not a contradiction; it’s a fairly accurate read of what’s happening. The same report notes that “businesses that own the agents will have a structural advantage in maintaining consumer relationships, driving full-journey engagement, and capturing data” — which is a polite way of saying the agent works for whoever deployed it, not necessarily for the person typing into it.

Agentic commerce is walking straight into that gap. Consumer surveys this year show 65 per cent of US shoppers trust AI to compare prices, but only 14 per cent trust it to actually place an order on their behalf. OpenAI paused its own Instant Checkout feature in March, shifting focus back to discovery rather than transactions — a fairly candid admission that the trust and evidence layer underneath these systems isn’t ready, even as the commerce layer races ahead.

Speed without an evidence trail is its own risk

It would be dishonest to hold vibe coding up as the clean counterexample to institutional advantage, because the same “move fast, verify later” instinct shows up there too, with its own cost. Security researchers who scanned over 1,400 production vibe-coded applications found 65 per cent had security issues and 58 per cent carried at least one critical vulnerability. Georgia Tech researchers tied 35 CVEs in a single month directly to AI-generated code. Building got easier. Building something that holds up under scrutiny did not.

Also Read: The hidden economics of autonomous AI agents

I think about this the way I was trained to think about a scene before I touch a patient: check the evidence, understand what’s actually in front of you, then act — not the other way round. That habit, more than any framework, is what shaped the signal-intelligence work I do now at OnTheRice, where the working rule is that a claim only counts once it’s timestamped, sourced, and checked against what actually happened later. I don’t say that to sell the product. I say it because the discipline is the point, and it’s transferable to anyone building in this window, not just to me.

What Southeast Asia’s builders should actually be arguing for

The uncomfortable version of this moment is that AI has quietly widened, not narrowed, the distance between who acts on information first and who reads about it after the fact. Capital in this region is barbelling into a handful of infrastructure mega-rounds while dedicated regional funds can’t close.

Consumer-facing AI is scaling in usage while shrinking in trust. And a genuine wave of non-traditional builders — paramedics, ex-guild leaders, people who never touched a computer science classroom — is proving the tools no longer gatekeep who can ship. That part is real, and it matters.

But shipping fast and shipping trustworthy are different achievements, and only one of them closes the actual gap. The founders in this region who deserve the next round of attention and capital are not the ones building faster checkout flows or flashier agents.

They’re the ones building the boring, auditable infrastructure that gives an ordinary person in Singapore, Jakarta, or Manila the same kind of advance notice a hedge fund buys for itself — evidence attached, timestamped, and checked against reality afterward. Everyone can build with AI now. The next competitive line is who’s willing to build something worth trusting first.

—

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 Everyone can build with AI now. Almost nobody can see what’s coming appeared first on e27.