01. The Core Feature of This AI Wave

The most important feature of this AI wave is that it makes scaling highly personalized services possible.

Over the past two years, the emergence of ChatGPT has ignited another AI boom. Despite much discussion about the AI application layer, no killer app has emerged yet.

Nevertheless, I firmly believe this AI wave is comparable to the Industrial Revolution because it will redefine supply-demand dynamics and create vast new possibilities. Its essence is a supply-side transformation of intellectual resources.

02. Supply-Side Transformation: Breaking the Scarcity of Intellectual Resources

In the past, intellectual resources were exclusively human and scarce — especially in fields requiring professional knowledge and skills, like education, psychological counseling, and legal services. The high cost of these services stemmed from their dependence on human intellectual labor, which is inherently limited.

Large language models are breaking this constraint.

If we look back at the Industrial Revolution, many "luxuries" once exclusive to the aristocracy became widely accessible:

Textiles: Before the Industrial Revolution, textiles relied on manual spinning and weaving — labor-intensive and expensive. Cotton cloth was even called "white gold."

Glassware: Glass-making was entirely dependent on manual glass-blowing, making glass windows and mirrors luxury items.

Books: Before steam-powered printing presses, book reproduction was slow and costly. Even with movable type, paper production remained expensive, and books were reserved for the wealthy.

Clocks: Only a few could own high-quality personal clocks. Ordinary people had no way to know the precise time — there was even a profession called "knocker-uppers."

Just as the Industrial Revolution made cotton, glass, books, and clocks accessible, AI is gradually breaking the scarcity of intellectual services, making "highly personalized services" scalable and accessible to ordinary people.

03. What Are Highly Personalized Services?

Here are some examples:

Emotional services: role-playing, emotional companionship, emotional counseling...

Professional services: legal consultation, psychological counseling, fortune-telling...

Life services: personal image design, interior design, career planning...

Educational consulting: personal tutors, professional skills coaching...

These scenarios share a common trait: they are highly dependent on intellectual resources, requiring a combination of professional knowledge and the user's unique background, and could previously only be provided by humans.

We chose to start with "fortune-telling." From the demand side, the need has existed since ancient times. From the supply side, quality practitioners have always been scarce.

In early 2023, when only GPT 3.5 was available, we built an AI + mysticism app called "Moonviz Tarot." Using purely AI-driven supply, it proved to be a fairly successful validation — both user reviews and payments were solid. With minimal investment, it nearly broke even through organic growth alone.

04. Why Can't Many AI Application Scenarios Work Yet?

AI applications face several fundamental challenges:

1. High customer acquisition costs: AI hasn't changed how users are acquired. Channels remain scarce and expensive.

2. New cost pressures: Large models bring new expenses (computing or token costs).

3. Technical complexity: Most scenarios require complex engineering to meet practical delivery standards.

The core variables are new experience and new cost. If (new experience - new cost) isn't significantly greater than (old experience - old cost), the scenario is hard to make work.

Building better new experiences takes time, and reducing new costs also takes time. Many scenarios face difficulties now, but that doesn't mean the difficulties can't be overcome.

05. From Scarcity to Accessibility

Despite encountering many difficulties in over a year of practice, I remain optimistic for the long term.

The cost of large models is dropping rapidly — a 75% reduction per year is normal. Each cost reduction makes more scenarios viable.

Just as the Industrial Revolution made clothing, shoes, and books — once scarce resources — accessible to everyone, the AI era will make "highly personalized services" across various domains scalable and part of ordinary people's lives.

06. Closing Thoughts

Entrepreneurship is tough, but I still often feel excited and fortunate.

After all, encountering such a wave, exploring and even creating interesting and valuable things in the process of era development — that itself is a wonderful journey.