It's an attempt at building a UI framework powered by an LLM, that is The LLM decides what to display, based on the following:
- Current State
- User Input
- Other Context
- We pass the UI Schema (ie: Available widgets) as context to the LLM
- We pass the current state as JSON
- The LLM is instructed to create a new state based on the input (Including the user instruction)
- Scalability, the more "widgets" we have the more context we need
TBD
make be # This will run the backend (Websocket server)
make fe # This will run the frontend (Static site, that connects to websocket)