If you haven’t heard, there is a AI race sweeping Silicon Valley.
While stalwarts like Google, Meta and Microsoft, and newcomers like OpenAI and Anthropic, grab headlines by unveiling model after model, many startups are trying to gain a foothold.
For investors, it can be difficult to tell which of these new entrants have real technical potential and which are likely to fade into technological oblivion.
This is partly why Jenny Xiaoco-founder of Leonis Capital, said it is important for venture capitalists to make a serious effort to understand the AI technology they are investing in.
Leonis Capital, founded in 2021, is rolling out its second fund of $40 million today. The venture capital firm focuses on next-generation AI companies and has invested in more than a dozen companies since its launch, including EnsureX, Motion and SpectroCloud.
Business Insider asked Xiao and his co-founder, Jay Zhao, to share the top five questions they ask founders when evaluating their technology. They shared their responses via email, which have been condensed and edited for clarity.
Xiao’s Top 5:
What becomes possible if the models improve by 10 to 20%?
The best founders don’t think in terms of incremental feature improvements, but rather in terms of capability thresholds. We want to know if they understand that advances in AI are often non-linear and if they can anticipate what future capabilities could fundamentally change or even break their product.
One of the most interesting takeaways from our Leonis AI 100 – in which we compared the most important AI startups – is that the strongest AI founders build just before the next technical breakthrough.
A good answer talks about entirely new workflows that are unlocked, not just marginal efficiencies, and clearly explains how the business would adapt or pivot as technology evolves.
Is what you’re building already on the roadmap for a foundation model lab?
Most AI startups fail not because they’re bad, but because they’re building something that OpenAI, Anthropic, or Google can eventually offer “for free” as a feature.
That’s why we don’t accept “they don’t focus on that” as an answer. We push the founders to explain what internal constraint would really prevent a foundation model laboratory to build the same thing. If the answer comes down to a question of focus or culture, it’s not a true divide.
Strong responses acknowledge that big labs could technically build it, but that doing so would break their incentive structure, pricing, or distribution model; require operational complexity that does not scale to them; or shift the value to downstream execution rather than model capability.
Some founders can also credibly claim that they have a 12-18 month head start. In contrast, weak answers rely on statements like being “more vertical,” having “more niche data,” or understanding customers better — responses we hear from about 95% of founders.
We’ve noticed a consistent trend: Most founders underestimate founding models, and most VCs underestimate them even more.
What data exists only because your product exists?
Ask “How much proprietary data do you?” is usually the wrong question, because no startup company has truly meaningful data. And if a founder’s pitch boils down to “we have more data, so our models are better,” that’s a red flag: Basic models often improve faster than proprietary models, and the companies building them have enough capital to buy data and quickly fill in the gaps.
What matters far more is not how much data exists today, but rather whether the product naturally generates better data over time. For example, industrial systems where data only appears once software is integrated into workflows, or products where change costs come from accumulated context.
What would happen if a competitor cloned your product within 30 days?
We ask this question to understand where defensibility really lies once code becomes commoditized. If a founder can’t answer it clearly, they probably don’t understand their own gaps.
Cosmetic benefits like code, UI, or templates are easy to copy, while structural benefits (like being a system of record or being integrated with compliance, audits, or standard operating procedures) are much harder to replicate.
The question also reveals founders’ temperaments: Strong founders are candid about their vulnerabilities (“Here’s what worries us the most”), while weaker ones tend to get defensive and insist that a threat “would never happen.”
What part of your system becomes more difficult to change as you scale?
This question requires founders to explain their early technical decisions and the compromises they intentionally made, revealing whether they understand the dynamics of the system rather than simply repeating features.
Strong answers sound like this: “We chose to execute actions, not just make suggestions, which increased accountability but created real lock-in” or “We became a system of record instead of a thin layer, which slowed integrations and sales at first but made organizational change costly later.” » In contrast, many founders default to saying they want to “stay flexible,” which often indicates they haven’t designed anything fundamental yet.
The best AI systems are designed from the ground up, deliberately removing degrees of freedom and hard-coded assumptions about workflow, authority, or data flow.
Zhao’s Top 5:
What did you have to unlearn to see this opportunity?
True understanding usually comes from letting go of old assumptions. This question helps distinguish founders who simply stumbled upon AI from those who experienced a real change in the way they see the world.
We are looking for intellectual flexibility, not degrees. Strong answers indicate a specific belief of the founder, why he believed it, and what made him change his mind; weak answers remain vague, like saying they “realized AI was going to be big.”
What would a well-funded team get wrong if they tried to copy you?
This question assumes that copying is possible and pushes the founders to explain what is not obvious. Shallow answers rely on statements like “they couldn’t move that fast” or “they don’t have our data,” while stronger ones point to hard-to-see benefits like deep operational insight, customer-specific integrations, or accumulated context that isn’t visible from the outside.
When was the last time you changed your mind about something important?
Founders who can’t point out when they’ve changed their minds generally don’t learn: they’re executing a fixed plan in a world that won’t stay that way.
This question reveals epistemic humility and asks whether a founder is truly seeking truth or simply seeking confirmation. AI moves too fast for people who can’t update, and we’re especially wary of founders who treat every pivot as a justification rather than a correction.
What needs to happen in the world – not just within your company – for this to work?
Market timing, regulatory changes, and platform dependencies kill far more startups than poor management. This question tests whether the founders think probabilistically about external forces and whether they designed to be resilient instead of relying on luck.
Founders who say “nothing” haven’t thought it through enough. Those we want to support can name several concrete external risks and explain how they protect themselves from them.
Tell me about a decision you made that those around you didn’t agree with. How did you experience this?
The best founders aren’t contrarians per se, but they can hold an unpopular position long enough for the world to catch up.