Advancements in Large Language Models (LLMs) have led to an increase in apps relying on APIs like OpenAI, Cohere, and Stable Diffusion. However, there are challenges with API-centered AI product development, including the lack of specialization, differentiation, and ownership. LLM APIs are general-purpose models trained on the whole internet, making it difficult to meet user expectations in narrow domains without additional manual checking. Additionally, customers may be paying for a polished prompt behind a pretty interface and reliance on APIs creates business risks. Ultimately, winners in the LLM app race will be those who quickly capture user traction, understand where the fundamental value lies, and build custom solutions that set them apart from the competition.

source update: The rise of API-powered NLP apps: Hype Cycle, or a New Disruptive Industry?