✨
Note: This content was translated from Traditional Chinese using Claude 3.7 Sonnet and revised by myself.
GPT-4o 的圖片生成模型幹翻了AI 新創：創業者該如何突圍？
OpenAI 的 GPT-4o 圖片生成技術讓許多 AI 新創面臨存亡危機。在巨頭不斷推進技術邊界的時代，創業者需重新思考定位與價值。成功的 AI 創業需跨越高門檻障礙、打造不可替代的使用體驗，並深耕專業領域知識，而非僅依賴簡單包裝的 AI 功能。
ReallyNiceDayHao Tien Cheng

When Giants Make Their Move, Where Do Startups Go?

OpenAI's new GPT-4o image generation technology has made many AI startups lose their reason to continue. This reminds me of an insight I saw earlier on Threads from Heptabase CEO Alan:

However, looking at AI development trends, we can observe the following three phenomena:

1. AI model capabilities are still improving very rapidly, while costs continue to decrease, with no signs of stagnation in the short term.
2. Major companies' General AI autonomous agents are continuously getting stronger (e.g., Claude Computer Use and ChatGPT Operator).
3. The range of Vertical Data accessible to major companies' General AI will become increasingly broad (e.g., Model Context Protocol).

If these three points continue to hold true, then two of the three competitive advantages of Vertical AI (data operations and specialized knowledge) will likely diminish to negligible levels in the near future, and possibly faster than we imagine. If General AI eventually performs as well as Vertical AI in every Vertical Task, what reason would we have to spend money purchasing Vertical AI for each Vertical?

The release of this new image model technology perfectly validates this concept. When OpenAI's 4,500 employees are at the cutting edge of technology, desperately thinking about how to make General AI better, what methods do you have to compete head-on with them?

Original post:

在 Threads 查看

The Pressure of General AI: The Dilemma of Competing with Tech Giants

Looking back, I had several ideas that were simply built on basic General AI packaging, including: HTML Email Generator, Anime-style Image Creator, and Interview Practice Tool.

The Limitations of Simple Packaging: Easily Replaceable Products

Each can essentially be generated with a single prompt:

  • "Create an HTML Email code for me"
  • "Create an xxx style image for me"
  • "Prepare an interview based on [CV.pdf] [JD.pdf], and act as a professional interviewer"

If we simply rely on such basic ideas to create products, we must be prepared to be crushed by General AI at any time. (Though I believe this is a way for modern junior developers to easily build a portfolio of projects.)

Strategies for Creating Real Value

If we want to create product value through Vertical AI, the most important thing is to find differentiation. The product you create should:

Overcome High Barriers

  • Does it overcome difficult barriers? Such as regulations, payment integration, etc. (these typically scare away potential competitors)

Build Irreplaceable User Experiences

  • Are the UI interface and user flow of significant value, allowing users to save time, create more benefits, and difficult to replace?

Develop Deep Domain Knowledge

  • Does your target market have unique professional knowledge or domain expertise that is difficult for outsiders to acquire?

Case Study Analysis

Stripe: Overcoming Real Pain Points

Before Stripe launched, over 1,000 developers knew existing solutions were terrible but took no action

  • Secured contracts with banks
  • Required knowledge of intricate credit card payment structures and details
  • These barriers scared everyone away
Ideas that require crossing seemingly difficult barriers are very likely to succeed. Boring ideas also have a higher success rate than interesting ones.

Valtec AI: The Unique Value of Expertise

Valtec AI leverages the founder's Air Force drone technology experience, integrating AI with various sensors to significantly improve the safety, efficiency, and sustainability of deep-sea fishing (I really like how minimalist their homepage is, lol)

Conclusion: The Path to Entrepreneurship in the AI Era

In an era where OpenAI, Google, and other tech giants are constantly pushing the boundaries of AI technology, entrepreneurs need to rethink our positioning and value. Successful entrepreneurship requires deep domain expertise, solving real pain points, and building moats that are difficult to cross.

Stripe's success isn't just because they had better payment technology, but because they solved complex problems that others were unwilling to tackle. Valtec AI's value comes not just from AI models, but from the founder's unique professional background and industry insights.

True innovation isn't about using the latest models, but finding unique ways to combine technology with deep domain understanding to create irreplaceable value. This path may be difficult, but it's the only way to truly establish yourself in the AI wave.