The Tangent thesis

14/6/2026 ☼ tangent ☼ incubator ☼ entrepreneurship ☼ VC

tl;dr: Tangent is a new incubator in Singapore I’ve just started. It is deliberately not a venture capital incubator. Instead, Tangent looks for ideas that solve real problems in existing markets using technology that already works, that take the existence of AI into account, and that can become rapidly self-sustaining. Because Tangent isn’t chasing venture-scale outcomes, it can widen the funnel to a much broader range of people and ideas than venture capital can. Good knowledge of real problems is distributed far more widely than capital and connections are.

For the last few months I’ve been building Tangent, a new incubator based in Singapore. I’ll be writing a lot about it in the months to come. The part to explain first is the logic — Tangent’s underlying thesis — because it’s the unusual bit.

Tangent is not a venture capital incubator. In some ways it’s the opposite.

Venture capital is built to fund a particular kind of idea: an enormous but not-yet-existent market that has to be created at great expense, and/or new technology that has to be invented and validated under conditions of great uncertainty. That’s what VC is for.

Tangent looks for the opposite: Ideas that solve real problems in existing markets by applying technology that already exists.

What follows is the three questions we use to decide what to back and who to work with. Answered the way we answer them, they also widen the range of people and ideas we get to support.

A real problem for an existing market, solved with existing technology

The first question: Is this an idea that solves a real problem for an existing market, using technology that already works?

We’re largely uninterested in markets that don’t exist yet and technology that hasn’t been invented yet. What interests us is knowledge of context: understanding a specific problem, the specific people who have that problem, and how to solve it in a way that visibly and immediately works. The starting point is always a real problem, a real population of people with that problem, and existing technology to address it. This applies as much to a not-for-profit or a social enterprise as to a business.

Informed by AI

The second question: How does this idea take the existence of AI into account?

There are at least four ways AI can shape a business. Type 1 is building AI itself: the models, or the hardware used to train them. Type 2 is building things that use AI to operate, calling a model every time someone uses them. Type 3 is using AI to build things that then run without it: AI helps make the product, but the product itself needs no AI. Type 4 is building things that are necessary or important only because AI is now in the world.

Our primary focus is Type 3. The cost of building software has been falling for years, but AI-enabled code generation made it drop sharply. It’s now possible to build narrowly scoped software much faster and more cheaply than before. That opens up something large and mostly unaddressed: boring tiny tools for individuals and small businesses it was never economic to build software for. (We’re interested in Types 2 and 4 too, but Type 3 is the centre of the thesis.)

Rapidly self-sustaining

The third question is the most important: Does the idea include a plausible case that it can become self-sustaining soon?

For-profit, social enterprise, not-for-profit; it doesn’t matter which. Every idea needs a credible path to the resources that let it keep going. Those might come from customers who pay, a government contract, a research grant to build a working prototype, or philanthropic funding paying on behalf of the people who actually use the thing.

The specific mechanism matters less than evidence that someone has thought about a mechanism for sustainability at all. We want people who can tell us what choices they’ll make (who they’ll go after first, how they might price, what they’ll build now and what they’ll defer) and why those are the right choices for their idea.

Widening the funnel on purpose

Asking and answering these three questions the way Tangent does leads to a valuable natural consequence. Because we’re not chasing venture-scale outcomes, we don’t have to select for the narrow profile that venture capital selects for. VC has to back ideas that could plausibly become enormous, which in practice means backing a specific kind of founder, with a specific kind of network, going after a specific kind of market.

Tangent’s thesis points us the other way. If the opportunity lies in boring tiny tools and other ideas informed by AI, then a wide range of small, context-specific ideas become viable. These ideas would never clear a VC’s bar for investment, but they’ll solve real-world problems for real people and can pay for themselves.

Good knowledge of real-world problems is distributed much more widely than capital and connections are. The person who understands a specific problem best is often nowhere near the usual entrepreneurship capital allocation funnel.

Tangent is designed to see what happens when we widen the funnel to consider people from a much broader range of socioeconomic backgrounds and ideas that respond to problems in a wider range of problem-spaces.

Because we’re trying something new, Tangent is designed to learn by doing. We can’t specify precisely the right programme in advance, so we don’t pretend to be more certain than we actually are. We’ll run a cohort and watch closely to see what works and what doesn’t, then evolve the programme. Expect Tangent to look different in a year, and different again the year after.

Join us

Our first cohort (13 July to 7 August) is full, but there’s still room in the cohorts running 31 August to 25 September and 19 October to 13 November this year. Dates for 2027 are posted too.

We’re looking for more people for those cohorts, and for guides willing to spend a two-hour block with participants. If you know someone who understands a real problem better than the usual funnel would ever find them, point them our way. Tangent’s website has the rest.

---

I’m the executive director of Tangent, a Singapore-based incubator of unconventional AI-informed startups, and a fellow of the Centre for Strategic Futures.

I help governments, foundations, companies, and startups design themselves to be better at doing new things, and have spent the last 15 years researching how organisations can succeed when things are uncertain. The Uncertainty Mindset is my book about how to design organisations that thrive in uncertainty and can clearly distinguish it from risk. I’ve also been working on tools that use uncertainty strategically, and idk is the first of these tools for productive discomfort.