Original · Unique Research · 2026-03-23 · Shanghai
Editor's note: This is the complete English rendition of Unique Research's March 23, 2026 interview-based article. First-person commentary belongs to the original author; investment judgments, views on AI and observations about founders are attributed to Piruze Sabuncu and the source. They are not independent investment recommendations, guarantees of business success, or experimental findings about what AI can or cannot do. The observations about Asian founders describe the interviewee's experience, not every founder from a region. The cited Canva, Airwallex and Supabase examples refer to Square Peg's institutional backing; the article does not establish that Piruze personally led each investment. Her biography and the firm's investment history have not been independently verified in this translation pass. All market and technology outlooks retain their historical March 2026 context.
Unique Awards · Guest Interview
Many AI Startups Ultimately Get Stuck Not on Technology, but on Never Truly Entering the Market
Rather Than Making a Product More Dazzling, Get It into the Market Earlier and More Deeply
Piruze Sabuncu | Investor at Square Peg
Focus areas: AI application layer & infrastructure layer
Representative companies backed: Canva, Airwallex, Supabase
When looking at AI startups today, it is easy to be drawn first to the technology.
Models are stronger, inference is faster, agents can do more, and product demonstrations are increasingly impressive. It is easy to develop an illusion: once you build the thing, everything else will naturally follow.
But that is not how the real world works.
After speaking with Square Peg investor Piruze Sabuncu, what stayed with me most was not which sectors she favored or her optimism about AI investing, but a more fundamental judgment:
Many AI startups ultimately get stuck not on technology, but on never truly entering the market.
Over the past two years, AI has lowered the barrier to "building a product" extraordinarily quickly. For the first time, many teams have capabilities they once could not have imagined: a few people can build something that looks complete, create a smooth experience, and even attract a respectable wave of attention on social media.
But something else has become harder:
Whether anyone actually needs what you have built on an ongoing basis.
In the interview, Piruze said one of the most common mistakes she sees is that founders fall too easily in love with their own technology, instead of keeping their attention on the customer's problem. Technology matters, of course, but it is a means, not the destination. What really determines whether a startup can move forward is whether it solves a user's problem so well that the user feels they cannot do without it.
That is also why many AI projects look decent at first glance—even quite good—but do not go far.
It is not because they cannot be used.
It is because they have not genuinely become embedded in a specific, stable, recurring scenario.
Not "Can It Be Built?" but "Does It Need to Exist?"
In many earlier waves of entrepreneurship, technology was the most obvious barrier. If you could not build something, you simply could not build it.
The AI era is a little different.
Today, it is more common for everyone to be able to build something. The question is not "Do we have the ability to launch this feature?" but "Once it launches, can it actually stick around?"
Put plainly, one of the easiest illusions in AI entrepreneurship is:
Mistaking "can build" for "should build."
These projects tend to share several characteristics.
They all look decent, and some are quite clever. Their product pages are well made, their demos grab attention, and their feature descriptions are convincing. But ask a couple of deeper questions and you find the most important piece missing: they started from a powerful technology, not a pressing problem.
The difference is substantial.
If you start from a problem, you naturally ask who feels it most painfully, who needs it solved most urgently, who will pay, and who will keep returning.
If you start from technology, you are more likely to ask whether the model can improve a little more, whether accuracy can rise a little higher, or whether interaction can become a little smoother.
Once the direction is reversed, the more you do, the farther you drift.
Piruze said they repeatedly consider several things when evaluating a project: whether the founders are sufficiently candid, whether they truly understand the problem they want to solve, whether the market is large enough, and whether the business can establish lasting barriers to competition. What struck me most was not familiar VC vocabulary such as TAM or defensibility, but founder-market fit: whether the founder understands the problem in a way others do not.
That matters enormously.
In the AI era, technical capability is increasingly becoming a basic commodity. What separates companies is often not "Do you know how to use this technology?" but "Did you understand this problem earlier and more deeply than others?"
What Many Teams Underestimate Is Not the Product, but Distribution
Piruze made another important point in the interview: many AI founders underestimate GTM, or go-to-market. She was direct about it. Distribution remains one of the hardest problems in business, and AI does not automatically make it easier—especially in enterprises. No matter how good a product is, without a credible path into the market, it can easily stall.
That is worth unpacking.
Discussions of AI today readily focus on the product itself: how capable it is, how fast it is, how smooth the experience feels, and how intelligently it calls models.
But many companies' real difficulties occur outside the product.
For example:
Where will users come from?
Why would the first seed users stay?
Why would a customer trust you rather than wait a little longer?
Can the team turn a trial into steady repeat purchases?
Who inside an enterprise customer will drive adoption?
How many layers separate "this looks good" from "this gets a budget"?
These are the questions on which many AI companies actually get stuck.
Products have indeed become easier to build in the AI era, but that has not made markets easier to enter. On the contrary, greater supply fragments attention and leaves users with less patience. Whether a product can grow depends not merely on standout features, but on whether it can retain a place in real workflows.
Without entering the workflow, it will not become a habit.
Without becoming a habit, retention is difficult.
Without retention, growth is often little more than surface excitement.
What Is Truly Scarce Today Is Not Just a Good Product, but "a Product People Keep Choosing"
Many people now ask: with large models so powerful and major companies moving so quickly, do startups still have a chance?
Piruze did not fall back on an easy "opportunities are everywhere." Her judgment was more specific: there are opportunities in both applications and infrastructure, but the conditions are clear—you must serve a sufficiently large market and have the potential to build defensibility. Infrastructure must solve key bottlenecks in the AI stack. Applications face a harder question: how to build a defensible business in a real market.
What matters most here is not the standard answer that "both applications and infrastructure have opportunities," but the underlying logic:
What people lack today is not an AI product. It is an AI product they want to use repeatedly and that is not easily replaced.
That is why startups cannot build their narrative solely around "we are faster," "our model is stronger," or "our experience is smoother." Competitors can catch up easily, and these qualities may quickly become the industry's default.
The companies that endure still have to answer some very old questions:
Are you solving an essential need?
Is the group you serve sufficiently well defined?
Do you have your own way of acquiring users?
Why should users keep using you rather than switch after three weeks?
If a major company enters, can you still hold your ground?
These questions are not new. AI has simply made them important again.
Investors Increasingly Value Not Just Technical Judgment, but People
Piruze said it is very difficult to assess technical risk objectively in early-stage investing. Rather than pretend to predict precisely which technical approach will win, it is better to ask whether the founding team clearly understands the risks it faces, can navigate them, and is honest about what it does not know. She singled out self-awareness and intellectual honesty as often being as important as the technology itself.
That is interesting.
AI is so hot today that many people assume VCs care most about model understanding, technical judgment, and sector predictions. But her answers suggest that when it comes to actual decisions, people matter even more.
Especially in early-stage projects.
Technology changes, direction shifts, products are rewritten, and markets move. What cannot easily be copied is how a person understands problems, faces uncertainty, leads a team, and advances things step by step through disorder.
She expressed a similar view when discussing whether AI will replace investors. AI can integrate information, recognize patterns, and uncover public signals faster. But it cannot judge how a founder will behave at the hardest moment. Nor can it build the kind of trusting relationship that makes a founder call you first. The final investment decision remains a human one, simply supported more thoroughly by AI.
This is also a reminder to many founders today:
Do not understand fundraising as a one-off exercise in matching information.
It is more like the beginning of a long-term relationship.
Ultimately, investors are backing not just a market opportunity or a deck, but a team they are willing to accompany for many years.
Why More AI Companies Think About Global Markets from the Start
Piruze's own experience helps explain this. She grew up in Turkey, studied and worked in the United States, operated businesses in Asia, and now looks at projects across APAC. She noted that the best technology companies tend naturally to cross borders, as do the best founders. For her, one of AI's biggest opportunities lies at the intersections of geographic markets.
Behind this is a very practical change.
Many AI products today, from the day they are born, look less like the strongly regional internet products of the past. They are more naturally suited to global distribution and more easily compared worldwide. Teams may start in Australia, Singapore, or China, but their products and users will not necessarily stay local; the teams may later move with the center of their market. Square Peg's experience backing companies such as Canva, Airwallex, and Supabase reinforces its belief that truly ambitious companies may start in APAC, but usually do not target just one market.
So in AI entrepreneurship today, going overseas is no longer an optional move to "consider once we are big."
For many teams, it is built into the product logic from day one.
Of course, that also raises the bar. You face not just a larger market, but unfiltered global competition.
Where Many Asian Founders Lose Out Is Not Ability, but Being "Too Quiet"
One part of the interview struck me as particularly worth revisiting for Asian founders today.
Piruze said many Asian founders she has met are highly capable and genuinely work hard, but a common problem is that they keep their heads down and wait until a product launches or reaches a certain stage before telling their story. In the global startup environment, that habit can put them at a disadvantage. Investors, customers, and future employees often need to see, understand, and remember you before you officially launch. Her judgment was direct: narrative is not vanity, but a competitive asset.
I strongly agree.
Many founders still misunderstand "telling a story." They think it is the marketing team's job, and may even conflate it with packaging, exaggeration, or bluffing.
But a genuinely good narrative is not fabrication or the construction of a persona.
It is more like telling the market:
Why I noticed this problem.
Why I am better placed than others to solve it.
What exactly I want to change.
Why you should pay attention to me now.
This is especially important in AI entrepreneurship.
There are so many new products today, with feeds refreshing constantly. You have to do more than build something: you must also establish a point of reference in other people's minds. Otherwise, you may already be doing very well, but the market has never had the chance to get to know you.
Ultimately, AI Entrepreneurship Is Still a Very Traditional Business
Having read the full interview, my strongest impression was:
AI looks new, but many of the things that actually determine success or failure are old.
Are you solving a real problem?
Do you understand users?
Do you have a way to enter the market?
Can you find the right people?
Are you honest enough?
Can you build trust?
Can you find your place in global competition?
These ideas hold in any era. In the AI era, they have simply been amplified again.
Technology is changing so quickly that everyone tends to focus first on what is "new." Yet a company's ultimate survival often depends on things that are less new but remain effective.
So if I had to summarize Piruze's interview in one sentence, I would put it this way:
The first problem many AI startups really need to solve today is not how to make their product more dazzling, but how to get it into the market earlier and more deeply.
Technology determines how far you can go. The market determines how long you can stay.
Selected Q&A
Q: What does Square Peg value most when evaluating AI startups?
Piruze said they do not have a rigid checklist that applies to every project, but they repeatedly examine several things: whether founders are candid, whether there is founder-market fit, whether the market is large enough, and whether the business has the potential to build lasting defensibility.
Q: What mistakes does she most often see AI founders make?
Three are particularly typical. First, becoming so absorbed in the technology itself that they drift away from the customer's problem. Second, underestimating GTM and assuming a good product will naturally grow. Third, failing to recruit the right people early enough, leaving the organizational foundations of the first two years unstable.
Q: With large-model companies valued so highly, do startup teams still have opportunities?
Her view is not pessimistic. Both applications and infrastructure still offer opportunities, provided you solve a sufficiently important problem. Infrastructure must address key bottlenecks; applications must enter a large enough market with a chance to build a defensible business.
Q: How does Square Peg differ from VCs in the United States or China?
Its strength is not just the speed of funding, but long experience accompanying APAC founders as they grow across borders. It better understands the details, turning points, and complexities of entering global markets. Beyond evaluating companies, it also helps with GTM, fundraising, and key introductions.
Q: If a major company enters a vertical, is there still room for a startup?
She did not simplify the issue. Her focus is not pretending to predict technical approaches precisely, but whether teams truly understand the technical risks they face and have sufficiently clear judgment to handle them. What matters is not "Will there be pressure?" but "Do you know where the pressure will come from, and can you withstand it?"
Q: Will AI replace VC?
Her answer is no. AI can significantly improve the efficiency of research, analysis, search, and monitoring, but the final investment judgment remains human. Entrepreneurship is full of variables, and many critical moments depend not on public information, but on understanding people, building relationships, and accumulating trust.
Q: What particular reminder does she have for Asian founders?
One important point is not to wait too long to tell their story. Piruze believes many Asian founders keep their heads down too much and overlook the importance of establishing a presence in global markets. Narrative is not an optional bonus; it is part of competitiveness.
Q: What is the key to successful PLG?
She mentioned several elements: community, turning customers into advocates, getting influential people to share your best practices, and having the founder be the brand's most important marketer.
Q: Post-seed, which metrics do global VCs care more about?
Revenue and growth matter, of course, but are not enough. She places greater weight on underlying PMF signals such as cohort and retention, to see whether you are "genuinely retaining users" or whether "acquisition is merely temporarily outpacing churn." There is also something harder to quantify but more important: customer love. Would users genuinely be upset if your product disappeared? Would they proactively recommend it? These often reveal more than surface-level numbers.
This article was compiled from an in-depth interview with Square Peg investor Piruze Sabuncu.