Chat

Chat with an answer engine using RAG (Retrieval-Augmented Generation).

Path Parameters:

• engine_id: Unique identifier of the answer engine

Request Body:

• query: The user’s question (required)
• model_llm (optional): LLM model for generation. Default: llama-v3p3-70b-instruct
• mode (optional): vanilla or agentic. Default: vanilla
• top_k (optional): Number of context chunks (1-30). Default: 5
• conv_thread_id (optional): Pass previous thread ID to continue a conversation
• prompt_id (optional): Specific prompt ID to use for generation

Returns:

• answer_engine_id: ID of the engine that processed the request
• conv_thread_id: Thread identifier for follow-up questions
• query_id: Unique identifier for this query
• query: Echo of the submitted query
• model_llm: LLM model that produced the answer
• answer: Generated answer from the LLM
• token_usage: Token usage statistics

Raises:

• 404: Engine not found
• 400: Chat request failed

Example Request:

POST /api/v3/answer-engine/ae_8c625d9d71/chat
Headers: {"Authorization": "Bearer <api_key>"}

{
  "query": "What is machine learning?"
}

Example Response:

{
  "answer_engine_id": "ae_8c625d9d71",
  "conv_thread_id": "thread_9d9af946e7",
  "query_id": "query_4afc22621f",
  "query": "What is machine learning?",
  "model_llm": "llama-v3p3-70b-instruct",
  "answer": "Machine learning is a subset of artificial intelligence...",
  "token_usage": {
    "input_tokens": 312,
    "output_tokens": 124
  }
}