Don
Hoang

After 50+ investments across the full company lifecycle — from pre-seed to growth — I've come to believe that what makes a great founder changes dramatically depending on the stage. The framework I use today is deliberately different for early versus late.

Early stage: conviction over proof. At the beginning, there's no data worth trusting. What I'm looking for is a strong vision — not just a large market, but a specific and defensible point of view on how the world is going to change. The founders who stand out have a unique perspective that others haven't seen yet, or haven't taken seriously. That asymmetry is the whole game at the early stage.

Alongside that, I'm looking for genuine unfair advantage. Not just "we're working hard" — that's table stakes. I mean structural advantage: deep domain expertise, proprietary distribution, a network that can't be bought, or a technical insight that creates real distance from day one. The opportunity needs to be large, but the edge needs to be specific.

Late stage: vision paired with the science of execution. By the time a company is scaling, the romantic phase is over. I want to see the same strong vision and conviction — that never goes away — but now it has to be paired with rigorous operational execution. Metrics that compound. Clear accountability structures. A repeatable motion for growth that the founder can articulate and the team can run without them in the room.

The founders who struggle at this stage are usually the ones who thrived on instinct early and never built the systems to scale their judgment. The ones who thrive are the ones who treat execution as a discipline — who are as rigorous about performance data as they are passionate about the mission.

The best founders I've backed have both gears. They can zoom out to the vision and zoom in to the numbers — often in the same conversation.

There's a conversation happening in boardrooms right now that I find deeply misguided. It goes something like: "We need an AI strategy." What they usually mean is: we need to be seen doing something with AI before our competitors do.

That's not a strategy. That's anxiety dressed up as planning.

The operators I've seen use AI well — and I've been watching this closely across my portfolio — share one thing in common: they started with the problem, not the technology. They asked "where are we slowest, most error-prone, most reliant on manual judgement?" and then looked at whether AI could address that specifically.

At Revolut, speed of execution was always a competitive weapon. If AI had been as capable then as it is today, the places I'd have deployed it first wouldn't be customer-facing — they'd be internal. Compliance review. Risk modelling. Localisation. The unglamorous operational work that compounds over time.

The operators who will win the AI era aren't the ones who move fastest. They're the ones who make the most durable choices — who embed AI into their core operations in ways that are hard to unwind and impossible to replicate quickly. Infrastructure beats features, every time.

If you're thinking about where to start: find the workflow in your company that is highest-frequency, most rule-bound, and most expensive when it goes wrong. That's your first deployment. Everything else follows.

The most common mistake I see founders make with AI right now is treating it as a technology competition. It isn't. The underlying models are commoditising faster than anyone predicted. The moat is elsewhere.

What I'm watching closely as an investor: who owns the distribution, the data, and the workflow. The companies that will win aren't necessarily the ones with the best model. They're the ones embedded deeply enough in a customer's daily operations that switching becomes genuinely painful.

This is a pattern I saw play out at Uber. The technology was never the durable advantage — it was the network, the trust, and the habit. By the time competitors had comparable technology, Uber had years of behavioural data and market presence that couldn't be replicated quickly.

For AI companies building today: the question isn't "how good is your model?" It's "how deep is your integration?" Workflow depth creates switching costs. Switching costs create defensibility. That's where I'm looking.

The founders I'm backing in AI right now are the ones who can answer this question clearly: why will your customer still be using this in five years, even if a better model exists? If the answer is the model itself, that's not enough.

Most companies talk about moving fast. Revolut actually did it — and the difference between those two things is almost entirely about how you approach product execution.

When I was VP of Global Business, we were launching new products at a pace that most financial services companies couldn't comprehend. Not because we had more people, but because we had built a machine for iteration. The cycle from idea to live product was compressed to a degree that changed what was even possible strategically.

What made that velocity possible: ruthless prioritisation at the top, genuine autonomy at the team level, and a culture that treated a fast failure as strictly preferable to a slow one. We didn't launch perfect products. We launched products that were good enough to learn from — and then we iterated in public, quickly, based on real signal.

This is harder than it sounds. Most organisations slow down product development not because they lack talent, but because they lack clarity. Every layer of approval, every cross-functional dependency that isn't explicitly managed, every strategy that isn't specific enough to generate a clear "no" — all of it accumulates into drag. At scale, that drag becomes the defining constraint on what you can build.

The lesson I took from Revolut into how I work with founders today: product velocity isn't a function of effort. It's a function of organisational design. How decisions get made, how teams are structured around outcomes rather than functions, how you handle the tension between quality and speed — these are leadership choices, not engineering ones.

The companies that sustain high product velocity as they scale are the ones that treat it as an explicit priority — not an accident of a scrappy early team, but a system that's been deliberately built and protected as the company grows.

One of the most common questions I get from founders scaling internationally is some version of: how do you move fast in new markets without the whole thing falling apart? Uber's answer was radical decentralisation — and it's one of the most interesting organisational experiments I've been part of.

The model was simple in principle and brutal in execution. Each city operated almost like an independent startup. The city GM had real P&L ownership, real hiring authority, and real latitude to make decisions that fit their market. London wasn't run like São Paulo. Lagos wasn't run like Singapore. The assumption baked into the model was that the person closest to the market understood it better than anyone at HQ — and they were right.

What this unlocked was speed. When you don't have to escalate every decision up three layers of management, you can respond to a competitor move on a Tuesday and have a counter-offer live by Thursday. That kind of tempo is genuinely hard to compete with. Most large organisations structurally cannot move that fast, regardless of intent.

But decentralisation has a cost that's easy to underestimate. Without strong shared infrastructure — consistent data, shared playbooks, tight feedback loops back to the centre — you get fragmentation. Thirty-five cities re-solving the same problems independently. Institutional knowledge that stays local and never compounds. We learned that lesson in real time.

The version of decentralisation that actually works looks like this: maximum local autonomy on the decisions that are genuinely local, and maximum standardisation on everything that isn't. Pricing strategy in a new city? Local. How you track driver supply? Global. What counts as a successful launch? Global. How you build relationships with local regulators? Local.

The intensity of the culture at Uber in those years is hard to describe from the outside. It was genuinely high-pressure, often chaotic, and deeply competitive. But underneath that, there was a clarity of mission that kept people aligned across time zones and languages. Everyone knew what winning looked like. That shared definition of success was the invisible infrastructure that made the decentralised model function.

What I took from it: structure is a strategy. How you organise a company is not an HR question — it's a competitive weapon. The founders who treat it that way build organisations that can actually execute at scale.