AI entrepreneurs need not worry excessively about being "crushed" by giants like OpenAI. Although many AI startups rely on third-party foundation models, this "wrapping" strategy is common and necessary in the early stages. Many successful companies were initially built on third-party technologies, such as Salesforce on Oracle databases, and Box on AWS.
The key is to continuously add unique value on top of the foundation models, making the product more "substantial" through design, user interface, new features, and other means. New features from giants like OpenAI can also become new features for startups; the key is how to add sufficient additional value around these features.
Currently, many AI products are just at a "passing" level, with a significant gap to truly "excellent". While LLMs have lowered the barrier to building intelligent products, creating truly outstanding products remains very challenging, requiring deep understanding of customers, elegant design, and handling of various edge cases.
Startups should focus on solving specific problems for particular user groups, rather than trying to become general-purpose AI assistants. By deeply understanding user needs and continuously improving product experience, AI startups still have significant room for growth. It's important to avoid over-reliance on a single technology and instead build their own core competencies.