Bitcoin reaches a historic high, with Ether and a host of altcoins surging forward. Yet, I can't muster any interest; as we approach the peak of this cycle, I'm pondering what the 100x opportunity of the next decade will be.

And my answer is: AI and stablecoins

I have been increasingly feeling lately that the productivity paradigm represented by agents is profoundly changing our information, work, and life.

Later, it raised $15 million in investment from a16z at a valuation of $120 million.

Regardless of the quality of the product, I believe that this shift in direction is quite evident. The hiring requirements and work paradigms of tech giants like Google are clearly outdated.

Google was founded in 1998 and emphasized early on the importance of "smart engineers" and a solid foundation in computer science, particularly algorithms and data structures. Early interviews at Google would ask questions like "write a search algorithm with a time complexity of O(log n)". As Google achieved success, this hiring model was imitated by many startups.

However, as AI programming capabilities continue to strengthen, agents represented by cursor, claude code, and GPT are rapidly bridging this technological gap.

In fact, do we really need that algorithm knowledge? Actually, not really.

In a recent YC speech, Andrew Ng mentioned that in the past, the ratio of product managers to engineers in their team was about 1:6 or 1:7. However, recently, someone on their team proposed changing this ratio to 1:0.5.

This is because we now have the ability to quickly engineer and commercialize with the help of AI, while discovering and validating needs has become a scarce resource.

This also means that whether it is work or entrepreneurship, AI has given us a magical ability to quickly prototype and scale. The gap between everyone has turned into a matter of taste and feel.

A company that I have recently come to appreciate and learn from is Loveable, which is a perfect example of this.

This company is one that supports vibe coding, established at the end of 2023, and officially launched in November 2024. Just 8 months later (July 2025), it completed a $200 million Series A financing round, reaching a valuation of $1.8 billion, becoming one of the fastest-growing AI startups in Europe.

Their revenue growth is astonishingly impressive:

1⃣ Achieved 4 million USD ARR (Annual Recurring Revenue) within 4 weeks of launch;

Launched for 3 months, surpassed 17 million USD ARR, with 30,000 paying users.

Launched for 7 months, achieving 75 million USD ARR, with a total of 2.3 million registered users, of which 180,000 are paying users.

Behind it, besides hitting the windfall, a very important reason is that they are an "agent-native company".

Lovable has been an "agent-native" architecture since Day 1:

All workflows default to supporting AI agent invocation and collaboration;

The internal toolchain (CI/CD, deployment system, database management) also has an API interface for agents.

The work of the development team is designed around the concept of "What AI can accomplish, and humans only perform key checks."

Each agent has "work permissions", "contextual memory", and "communication interface", and can communicate directly through Slack or command line.

There is also an "agent performance review" internally (similar to an assessment), and if expectations are not met, the prompt may be replaced or adjusted.

I believe this represents a new company structure in the future, especially the best opportunities for small and medium-sized enterprises (even one-person companies).

This is also one of my focuses in the company, to achieve agentification as early and as much as possible.

Imagine if your daily information stream had AI organizing the sources, summarizing industry trends, and generating collaboration proposals (tob) for newly emerging opportunities based on a finely-tuned model from past cases, or automatically handling to-dos in work orders and customer inquiries (toc), and also conducting your own performance evaluations. Wouldn't that be a miracle?

In addition to work, in all aspects of life, travel, and trading, we can have an AI team to serve us. The efficiency brought about by this is a hundredfold increase.

In the future, we will no longer be engineers, analysts, and businesspeople, but rather project managers managing AI engineering teams, analysis teams, and business teams.

By fully committing now, we can gain a cognitive gap that is ahead of our time.

The next article will discuss stablecoins.