Conversation with "42ZhangJing" Qu Kai: The Earlier You Boldly Believe in AI, the Greater Your Potential Returns | AI Applications 100 Questions

"100 Questions on AI Applications" is an interview series by Linear Capital that explores the current state and trends of AI applications. We invite AI application entrepreneurs — founders from Linear Capital's portfolio or friends of the firm — to share their startup stories or personal reflections on the industry, discussing timely topics and progress in AI applications. We hope these interviews offer useful perspectives for others following this space.

In this edition, we invited Qu Kai, founder of 42 Zhangjing, to share his observations and takeaways from the past year in AI application entrepreneurship. 42 Zhangjing has published numerous in-depth interviews with AI application founders. Qu previously worked as a product manager at BAT companies and as an investor, and holds a degree from Duke University.

🎙️ Interview

1. Linear Capital: What's the most interesting AI application you've come across recently? What makes it stand out?

Qu Kai: The AI companion product EVE that recently blew up on Bilibili strikes me as particularly significant. EVE is a 3D AI companion, founded by Zhang Xiaofan, formerly CEO of Dreamscape Games. This is his second venture. Its first trailer dropped on Bilibili in mid-October and racked up over 1.1 million views in two days.

This is a polished, well-packaged product built by a professional game team that invested serious resources, combining established and emerging technologies while using AI to cover weaknesses and amplify strengths. For instance, they poured substantial effort into creating photorealistic 3D models on top of traditional companion product frameworks. When they hit the long-term memory limitations of existing models, they invented a 128-slot memory system to solve it. Ultimately, they used a complex engineering approach to crack numerous problems that standalone large models couldn't handle.

The emergence and viral success of this product tells us that the gap between AI technology and real-world application isn't as wide as people assume, nor are the obstacles insurmountable. What the market needs right now are professional teams that believe in this, plus more time and investment.

2. Linear Capital: Comparing 2023 and 2024 in AI applications, what notable shifts do you see in founders' chosen sectors, team product capabilities, and fundraising?

Qu Kai: Our sense is that fewer than 100 AI application companies raised new funding in 2024 (by this metric, the current market more closely resembles 2008–2009 — the underlying AI technology remains immature, and hasn't gained genuine capital market validation). Most of these were companies founded in 2023, with only a handful established in 2024. Of this new wave of AI startups across both years, roughly 20 have reached valuations above $50 million (counting only AI applications, excluding large models, embodied AI, and related fields).

I mentioned something on a 42 Zhangjing podcast: with such thin data, all analysis is essentially meaningless. Which AI sub-sectors hold the most promise? Which institutions are investing most aggressively in AI? What's the typical profile of an AI company founder? These questions all presuppose that AI itself is recognized as a major investable category — but actual institutional activity suggests otherwise.

That might sound pessimistic, but here's the counterpoint: among those ~20 companies valued above $50 million, the vast majority were founded in 2023. This illustrates something important — those who believed in AI earlier and more boldly have better odds of outsized returns.

3. Linear Capital: What do AI application founders worry about most? What do you hear most often as VCs' reasons for investing or passing?

Qu Kai: The founders I see are actually most preoccupied with model instability, hallucinations, costs — not fundraising. They've already accepted that funding isn't a prerequisite for starting up in today's market. Only by getting the model and product to work together can you find PMF and generate enough revenue to sustain yourself.

VCs can cite countless verbal or objective reasons for passing, but the real underlying reason is usually just one: in a fuzzy, rapidly shifting, highly uncertain market, they haven't figured out their own strategy. There are endless external factors and perfectly valid institutional reasons to decline, but the final decision to invest or not, and how, comes down to internal conviction.

4. Linear Capital: After talking to so many AI entrepreneurs, what consensus within the founder community remains poorly understood by outsiders?

Qu Kai:

1. Commercialization has never ranked this high in priority

2. The importance of building model and product evaluation frameworks

3. High-quality data sometimes matters more than model capability

4. Engineering difficulty and moats exceed what most people imagine

5. Finally — not necessarily consensus, but certainly widespread preoccupation: whether to go global, and how, at the product, company, and team levels

5. Linear Capital: In AI entrepreneurship broadly, do you see more people with hammers looking for nails, or nails looking for hammers? If there's a clear tilt, what's behind it?

Qu Kai: Definitely more hammers seeking nails, which I think is perfectly defensible. Major opportunities always emerge from change. The nail is constant; the hammer is what changes.

6. Linear Capital: How do founders view moats in AI applications, and how do they think about building them?

Qu Kai: I genuinely believe moats ultimately come from the team, from insight, from the founder's decision after decision. I struggle to imagine how early-stage ByteDance, Pinduoduo, or Xiaohongshu could have articulated a definitively valid moat.

When OpenAI launched GPTs, many saw it as a devastating blow to products like Dify.ai. But Dify's founder rapidly repositioned the product to become the go-to choice for enterprises deploying their own internal GPTs, turning the threat into faster growth — truly converting crisis into opportunity.

7. Linear Capital: What drives founders to start companies in today's cold funding climate? The obvious answer is not wanting to miss the AI wave. But on the application side, the wind hasn't really picked up yet. What's the prevailing mentality — optimistic, pessimistic, or something else?

Qu Kai: I think more people are in a wait-and-see mode. The most optimistic, adventurous types mostly jumped in last year. Those remaining likely need greater certainty, more evidence.

But we've recently identified a valuable trend — a demographic we've termed "AI second-generation": people who spent the past year or two cutting their teeth at this wave's AI companies and products, now starting their own ventures. We're very bullish on this group's potential.

8. Linear Capital: What's the biggest change AI has brought? People have long talked about AI-native applications, about every app deserving to be rebuilt. Does that framing still hold, or have more promising niche scenarios already emerged?

Qu Kai: On this I'm still optimistic. We're seeing models getting stronger and cheaper, deeper collective understanding and methodology for building products around models, and numerous products that have found PMF.

So I still believe AI bots will substantially replace SaaS. I believe ToC applications will take new forms. I believe that as language, image, video, audio, and 3D models continue evolving, products will emerge that completely exceed our imagination. On many fronts, we no longer need qualitative leaps — we need quantitative accumulation.

9. Linear Capital: Based on your observations, what traits define founders who will succeed in this cycle?

Qu Kai:

1) The brave

They dive in when markets and technology remain uncertain. The insight they accumulate through this process is immense wealth.

2) Fast learners

The pace of model and technology evolution exceeds most people's imagination. Image generation techniques and papers may shift monthly or even weekly — from ControlNet to InstantID to AnimateDiff, from Midjourney to Stable Diffusion to Flux. You genuinely may no longer be able to tell whether a Xiaohongshu beauty influencer account is human or AI. Only fast learners can seize the latest opportunities as technology advances.

3) Strong executors

We want to see more founders who rapidly translate new insights and technologies into deployed solutions. The faster markets change, the more execution matters.

4) People skilled at applying technology to solve problems

Models, algorithms, and technologies are ultimately a founder's arsenal. The crucial ability is matching the technologies everyone sees to the right domains most precisely, solving the most valuable user needs. This is why we believe product managers still hold enormous value in the AI era.

5) People with commercial acumen

Good founders need to confront today's market reality objectively. This isn't the timing of Douyin, or even Toutiao. The priority is building this era's Neihan Duanzi first — making money to keep the team alive is the foundation for long-term growth.

Finally, though I've listed all these points, everyone's definition of an excellent founder differs, and there's never just one archetype. But I don't believe this changes over time or varies by cycle. Ultimately, exceptional people are those who truly transcend cycles and withstand the test of time.

10. Linear Capital: Have you seen any best practices for monetizing AI applications?

Qu Kai: The best-monetizing AI applications currently tend to be tool-oriented products, especially those going global.

11. Linear Capital: Any advice for AI application founders?

Qu Kai:

Stay optimistic in the face of massive uncertainty.

Keep dreaming while keeping your feet on the ground and making money.

📮 Further Reading

Linear Bolt Bolt is Linear Capital's dedicated investment program for early-stage, globally oriented AI applications. It carries forward Linear's investment philosophy, focusing on technology-driven transformative projects, and aims to help founders find the shortest path to their goals — whether in speed of action or investment mechanics. Bolt's commitment is lighter, faster, and more flexible. In the first half of 2024, Bolt invested in seven AI application projects including Final Round, Xinguang, Cathoven, Xbuddy, and Midreal.