Question about Supabase Vector Store

Does OpenAI/Gemini utilize function calling internally?

Possibly, if you're utilizing the callin.io AI Agent with semantic search capabilities or integration with LLMs. Many current RAG (Retrieval-Augmented Generation) workflows use:

• Supabase Vector Store to store embeddings.
• OpenAI function calling or similar to invoke functions like “searchdocuments” or “getcontext”.

However, this depends on the design of the AI Agent. If you're using the callin.io AI Agent with query: function, it might be relying on LLMs to generate functions based on the context.

How does the agent determine function parameters?

This implies that the AI Agent operates implicitly (leveraging LLM context) rather than declaratively, as seen in OpenAI function calling with a defined schema. Therefore, yes, you need to assist the LLM.

Should I include the schema in the system prompt?

Yes, it is advisable. If you do not explicitly define the types and columns in the function definition or metadata, you must provide:

• Table name
• Relevant columns (names + data type)
• Semantic context, if applicable.

How is data mapping handled during insertion?

Within the context of callin.io + Supabase Vector Store: Text is converted into embeddings (using OpenAI, etc.).

The resulting vector embedding, along with its metadata (original text, IDs, etc.), is inserted as a row into a Supabase table. callin.io allows you to map which fields are saved via the node fields.

You must ensure that the table possesses a vector column type and that the node is configured with the appropriate embedding model.

This discussion was automatically closed 7 days following the last response. New replies are no longer permitted.