After visiting AI entrepreneurs in Shanghai and Shenzhen, the author found that the truly successful products all share one common trait: an unwavering focus on a single direction, with relentless attention to detail. The author believes that many entrepreneurs today aren’t bottlenecked by “not enough AI,” but by insufficient granularity in execution and weak foundational skills. While AI-native approaches offer rapid and flexible development capabilities, the core traits required to be a successful entrepreneur—patience, perseverance, and steady, methodical effort—have not changed. The author emphasizes that an entrepreneur’s own “model” cannot rely on AI; it must be built through personal experience and continuous learning. The key to solving problems lies in focus and meticulousness—these remain the eternal core competencies of entrepreneurs.Article author and source: GeekPark
Last month, I made a focused trip to Shanghai and Shenzhen and had several in-depth conversations with entrepreneurs actively working at the forefront of AI.
After working on his new product for six months, I asked Qi Junyuan what his biggest takeaway had been so far. He said something simple but profound: “When you actually roll up your sleeves and get started, you realize how important it is to be serious about it.”
This remark sounded ordinary at the time, but it kept coming back to me later—and viewed in today’s industry context, it’s quite sobering.
I’ve had in-depth conversations with at least 200 AI entrepreneurs over the past three years, and now some are beginning to feel discouraged. The notion that “AI applications are no longer investable” is spreading within the industry—people feel this can’t be done, that can’t be done, that opportunities are gone and the window has closed. But if you look at it from the opposite angle, every truly successful product shares one common trait: they firmly commit to a single direction and obsessively refine every detail.
For example, with Junyuan’s product, what truly amazed me wasn’t the innovation in its form, but its data performance far exceeding my expectations. When I asked him how he achieved it, he said it came down to meticulous attention to detail—optimizing data, frameworks, and engineering—and next, he plans to focus on post-training, all of which involve painstaking, granular work. Step by step, inch by inch.
The same goes for Liu Yu from Vivix. He showed me the generation efficiency of their model—about nine audio-visual synchronized videos in 10 seconds. Three years ago, when Demi and his team were working on Pika, producing four videos in three minutes was considered incredibly fast. Personally, on Jiemeng, I usually wait for ages to generate just one video, so seeing their performance of nine videos in 10 seconds was truly surprising. When I asked how they did it, he said the model’s objectives were different, so the architecture was different. But what really mattered was that they had to make engineering optimizations at around 40+ points—each small improvement, when combined, resulted in over a hundredfold performance gain, while also significantly reducing costs.
Nothing new—just attention to detail, persistence, and seriousness. Slowly and carefully, bit by bit, small wins add up to big victories.
So I’ve been thinking: the core of an AI-native entrepreneur may not lie in “AI-native,” but in “entrepreneur.”
Many entrepreneurs in today’s industry face bottlenecks not because they lack AI or because there are no opportunities in this era, but because their execution is too shallow and their foundational skills are lacking.
What AI-native brings is speed, minimal personnel, low capital requirements, and strong building capabilities—this is the "spirit" aspect. AI offers us more opportunities, but also makes it easier to lose focus.
But the core qualities required to become a qualified entrepreneur have not changed in this era. You need a solid worldview, a clear path design, patience after setting goals, determination, and the ability to build strong positions and fight steady, methodical battles. This is the part of “simplicity.”
The issue lies here: there’s conflict between “spirit” and “deliberateness.” When you’re accustomed to the fast approach and can quickly reach 60 points, achieving 80 or 90 points might take ten times as long. Then you find that tenfold time investment unbearable and give up.
This is a problem.
A friend who works as a go-to-market service provider also mentioned to me that many founders aren’t skilled at marketing themselves but want to be hands-off, assuming that simply paying a vendor will deliver results. In the early stages of entrepreneurship, your product is still evolving and may need to pivot—you need to use this time to deeply understand the market and feed insights back into your product. During this phase, marketing isn’t about achieving maximum impact; it’s about maximizing efficiency. If you simply spend money or hire someone to handle it for you, you’re wasting the opportunity to personally learn and experience every aspect of GTM. Today, marketing is a critical component of competitiveness—if you don’t know how to do it and aren’t willing to invest real effort, the outcome won’t be good.
In short, it’s still the same issue: not taking it seriously and not spending time on what you’re not good at.
For example, WorkBuddy isn’t the first to pioneer its product form, nor is it the most advanced. Many tech enthusiasts might criticize it: “No real technological breakthroughs,” “It locks down too much freedom,” “Too simplistic.” But look at the target audience it serves and the value it delivers—its understanding of users, value proposition, UI, interaction design, gamification, and operational strategies are all meticulously refined. The subtle, proven tactics that Tencent once used for consumer products are applied here with exceptional smoothness.
These aren't innovations born from meetings or brought about by AI; they are the fundamental skills accumulated over many years and deeply rooted in the organization.
So I’d say that today, the fundamental skills of entrepreneurs are actually in short supply. The AI-native approach can be both learned and applied, and it’s relatively straightforward. What truly requires cultivation and practice are the core attributes of entrepreneurship itself.
Your product may be able to leverage the world’s best models, but entrepreneurs’ own “models” have no external API—they can only rely on the accumulation of historical data and the efficiency of real-time reinforcement learning.
The environment is always changing, but it’s precisely in this change that the true value of the word “entrepreneur” shines. The inherent strength of entrepreneurs is simple: as long as the problem you’re solving is real, and you stay focused and meticulous, you’ll always have an opportunity.