Unit of value of an AI platform

Earlier this year I wrote an article describing how OpenAI created a platform with the help of plugins—and APIs.

We’ll transition into a stage where API products will be focused on single features so chat users can easily consume them.

The focus here is “on single features.” Why?

AI systems that are open have to inherently be able to connect to other systems to perform actions. By contrast, closed AI systems offer access to actions as internal features.

OpenAI, for instance, can’t perform any actions on its own, other than the ones related to language understanding and manipulation. Try, for instance, to ask ChatGPT to make a drawing of a house. It will tell you it can’t draw things because it’s a text-based model. More sophisticated actions are even harder for the system to perform, so it relies on external systems.

The unit of value of an AI platform is a single action. The AI platform acts as an aggregator of actions. It can pick which actions to perform based on the perceived intent of the user it interacts with.

To thrive in this AI-oriented system APIs have to adapt and offer one feature with very high quality.