Instead of using LLM, can we use an API?

[FaresKi]

Hi!
Generative UI with Vercel AI looks really cool. I'd like to explain my use case. I'm currently using OpenAI w/ LlamaIndex to provide a RAGified endpoint with which users can interact in a scoped manner.
I've built an API that I've used for a chat interface, using ai/react's useChat method. This method sends and gets messages from this API (similar to this)
I'd like to know whether it's possible (or will be possible), to reuse the same concept as an API instead of the native LLM directly?
Thanks a lot!

[MaxLeiter]

Yep. The render helper relies on an LLM call, but you can use the streamable primitives with any async logic, not just LLM calls. Look at the example here: https://sdk.vercel.ai/docs/api-reference/generative-ui/create-streamable-ui