Eligible for binding votes: @daixiang0, @yaron2, @artursouza, @mukundansundar, @JoshVanL

Proposal looks good. Since a main benefit of this API lies in its abstraction layer, it would be good to have at least two components to begin with for the Alpha stage. Which components would be a good fit for the first stage in your opinion?

OpenAI is good for start, most LLM support it.

It would also be good to investigate how we can integrate with LangChain and offer higher level integrations for it with existing APIs like state, pub/sub and workflows in addition to providing native Dapr features like authentication, authorization and resiliency policies.

This article may help us think better to integrate Conversation API with existing APIs. Also, it shows an issue -- the existing LLM framework is not good for startups to quickly iterate, which is our chance.

The first step we can easily move is integrate Crypto API to use LLM API keys.

+1 binding

@artursouza has moved those into metadata and update the HTTP example.

I changed the proposal to move the metadata to component. I also added conversationContext to allow tracking of the chat and recover a session.

@artursouza thank you very much!

+1 binding