Three rules of AI achievements I learned at a top - tier AI conference in Europe.

God Translation Bureau is a translation team under 36Kr, focusing on technology, business, workplace, life and other fields, and mainly introducing new technologies, new ideas and new trends from abroad.

Editor's note: Stop building the "Louvre" in the AI world. The practice of Chinese teams has proven that the speed of solving real problems is the deepest moat. This article is from a compilation.

When I stepped into the Louvre to attend RAISE, the largest AI conference in Europe, I felt that the choice of the venue itself was sending a message: European artificial intelligence should be regarded as a masterpiece on par with the greatest achievements in history. The French really know how to create a grand occasion.

"The brilliant lines form words. They first illuminate themselves and then the walls. The combination of stones and words here constitutes both a wall and a text."

I have no intention of questioning this ambition. But after two days of discussions, roadshows and social events with champagne, my clearest insight was not about the rise of European AI. Instead, some deeper truths about AI impressed me: speed beats perfection, focus beats features, and domain expertise is the only way to compete with traditional technical backgrounds.

Focus Brings Speed

The most fascinating story comes from Tao Zhang, the Chief Product Officer and co-founder of Monica. Monica is a Chinese team based in Singapore, and they are developing a tool called Manus, aiming to enable anyone to independently complete complex real-world tasks through AI.

Tao Zhang initially noticed that people were constantly copying and pasting inside and outside ChatGPT to add context or export answers. This was just one of the pain points. They developed a Chrome extension for this and it was well-received by users. Then they quickly took the obvious step: developing a full-fledged browser. But instead of immediately launching the product, they conducted pious internal testing (dogfooding). After using their own browser day after day, they felt something was wrong. They watched AI process tasks but never knew when they would be completed, and realized that doing it themselves might be faster.

Meanwhile, Tao Zhang observed that the AI programming assistant Cursor was experiencing explosive growth. Even his wife, who is not a software engineer, was using it for data analysis - she just needed to enter her requirements in the chat box and click "Accept All" for any generated code. At that moment, he had an epiphany: Forget about the browser and build a Cursor-like product for everyone.

This transformation took only a few weeks, not months; the browser development took seven months, while Manus only took two. When an engineer asked if a model selector should be added, Tao Zhang refused: ordinary users don't care about models, they only care about results. "We should be experts," he said.

This combination of speed and focus is everywhere. The United States has adopted a lax regulatory approach to AI, which allows companies to freely invest, train models and develop products without dealing with red tape or bureaucratic procedures. This will of course bring problems - for example, xAI's Grok claimed to love Hitler and launched a female anime AI companion on the same day the company won a $200 million contract from the US Department of Defense. At the very least, the situation was quite embarrassing. The company won the contract thanks to the capabilities of its model. Considering that they went from scratch to the top of the industry in just two years, you can't deny their development speed: AI-native applications like Windsurf and Cursor are growing their revenues faster than any software category in history.

While pioneers in the United States and Asia are creating record revenues with huge investments in infrastructure, in Europe with a strict regulatory environment, people are still discussing how to obtain GPU resources.

The market's verdict is clear: either act quickly to solve real problems or be overtaken by others. Manus understands this well. Its product roadmap is highly focused on extending the independent working hours of AI - from two hours today to 24 hours in the near future - and is committed to making this ability accessible to everyone.

This is the winning formula in the AI field. Your first idea is probably not the best, but you can only discover this when you act quickly enough to try multiple methods while remaining focused enough to identify viable solutions. Manus followed this strategy. After launching the product in March and attracting a waiting list of 3.5 million people, it successfully raised $50 million in financing and its valuation quintupled.

On the right is Tao Zhang, co-founder of Manus

The New Definition of "Technical Founders"

Every venture capitalist I heard at RAISE mentioned a similar point: the era of non-technical founders is over. However, there are some important complexities behind this statement.

Tao Zhang and his other two co-founders of Manus know how to code, but their real advantage lies in their ability to dive deep into implementation details and then immediately step back to the macro perspective of user problems. When they decided to abandon the browser project, they didn't need three meetings or technical advisors. They just knew intuitively.

Alexandra Mysoor, the CEO of Alix, chose a different path. Her company uses AI to simplify the estate settlement process. She spent more than two years understanding every pain point families face in this process. This expertise allows her to distinguish which AI applications are truly important and which ones just sound clever but solve no problems. When her team develops features, she can immediately judge whether they are effective because she has experienced these problems firsthand.

The founding team of DeepSeek comes from the financial world, not the traditional AI research field - they were quantitative traders. Their advantage lies in understanding complex systems and having the ability to directly build their intuitions. Whether it's programming, law, medicine or teaching, the real meaning of "technology" today lies in whether you can personally verify whether a solution is effective. So, in a sense, "non-technical" founders are indeed outdated. But from another perspective, AI also allows more founders to be considered "technical".

Your Unfair Leading Advantage

This speed advantage also compounds in another way, and most American technology experts are not fully aware of this. English has always been the lingua franca of large technology companies, but when the technology itself is language, the situation is very different.

Given the prevalence of English on the Internet and the vast amount of data, language models perform best in English and will continue to generate more and better English data. English is the "reserve currency" of AI. Just as the global financial system operates on the basis of the US dollar regardless of your local currency, the global AI also operates largely on the basis of English regardless of your mother tongue.

For American entrepreneurs, this is like the compound interest you're already earning. Your default language is the default language of AI, and your first batch of customers expect products to have AI-native features because OpenAI has opened up the market. This advantage is not permanent - China is developing Chinese-first models, and Europe will eventually build its infrastructure. But before the rules of the game change, this advantage is worth taking good advantage of.

Observations in the Louvre

The French really know how to organize events - there's champagne at every break, authentic Italian espresso is provided, and the participants are very diverse. The European AI circle may not be very large - they call a €50 million financing a "huge financing", while American companies can get a $100 million seed round. But there are also some bright spots. As one of the most prominent AI success stories in Europe, Mistral competes by open-sourcing models and selling consulting services, proving that you can still build something substantial by adapting to local constraints.

Nevertheless, there are many similarities on both sides of the Atlantic. The same venture capital firms are writing the checks - investors from Sequoia and Andreessen Horowitz are everywhere. People also generally admit that most of the growth capital comes from American investors.

Standing in that ancient and world-famous museum, surrounded by the crystallization of human wisdom, these similarities become obvious. The craftsmen who built the Louvre mastered the most powerful tools of their time. And the most powerful tools today reward the same qualities as they did back then.

However, today, geographical location is less important than the basic laws of how things develop. Act fast enough to test multiple ideas; stay focused enough to solve real problems; build where the ecosystem advantages can bring you compound interest. And remember - the entire industry landscape and where the opportunities lie are changing rapidly, so take advantage of them while you can.

Translator: boxi.