RAG Query Endpoint

Executes a retrieval augmented generation (RAG) query on a dataset.

Request Payload:

• dataset_id: The dataset ID to query.
• messages: A list of messages that form the conversation. Each message must have a sender field (either ‘assistant’ or ‘user’) and a content field. The conversation must end with a user message.
• top_k: The number of documents to retrieve from the underlying search.
• keyword_column: (Optional) If provided (and not null), a hybrid search is used based on the specified column. If omitted or null, a vector search is performed.