GPT-powered startups face three major obstacles to achieve long-term success: Productivity Enhancements, Non zero-sum-game Value, and Moat = Value from Context. By analyzing their use case along these dimensions, startups can be classified into three levels of potential success. Level I applications focus on increasing productivity, while Level II applications provide value outside of existing zero-sum games. Level III applications leverage unique features such as in-context collaboration, gated knowledge/data, and offline capabilities to build a moat against generic GPT solutions. However, level I applications may face challenges in a crowded market of similar competitors. Real-world applications and a focus on long-term value creation may offer more defensibility. Overall, this framework provides a guide for CEOs, investors, and R&D budget responsibility to evaluate the potential of GPT-powered initiatives.
| Signal | Change | 10y horizon | Driving force |
|---|---|---|---|
| Three-Hills Model for evaluating GPT startups | Evaluating GPT startups | In 10 years, GPT startups will be evaluated based on their ability to surpass “Moat Mountain” | Identifying successful GPT startups |
| GPT-powered startups disrupting white collar industries | Disruption of white collar industries | Most jobs will become assisted jobs by the end of 2024 | GPT technologies and their impact on industries |
| Productivity Hill - GPT applications increasing productivity | Productivity enhancement through GPT | GPT applications will help users perform tasks faster and more efficiently | Speed increase and democratization of abilities |
| Tug-of-War Valley - Challenges faced by Level I GPT applications | Competition and counteraction in post-GPT world | Level I GPT applications face challenges from opposing GPT functionalities | Balancing value creation with opposing forces |
| Value Peak - Companies providing value outside zero-sum games | Value creation beyond zero-sum games | Companies providing value beyond human-centered tasks will reach Value Peak | Value creation in personalized experiences and problem-solving |
| Displacement Canyon - Competition with generic GPT solutions | Competition with generic GPT applications | Companies need to differentiate from generic GPT solutions to succeed | Competing with generic GPT assistants |
| Moat Mountain - Building a moat against generic GPT applications | Building a moat against generic GPT applications | Companies need in-context features, gated knowledge/data, or edge computing to defend against generic GPT applications | In-context collaboration, gated knowledge/data, and edge computing as differentiators |