RAG & Retrieval

Super-Rag with SDK

Superagent also supports using SuperRag via SDK. Let’s see how to use it.

Ingesting data to Vector Database

You can do this by sending a POST request to the /ingest endpoint. Here’s an example.

Plugging SuperRag into Superagent

To use SuperRag, you need to add it to Superagent as a tool.

1tool = client.tool.create(request={
2    "name": "Financial Reports data", 
3    "description": "Useful for answering questions about financial reports", 
4    "type": "SUPERRAG",
5    "metadata": {
6        "vector_database": {
7          "type": "pinecone", # make sure this is the same as the type of vector database you have used in SuperRag
8        },
9        "index_name": "financial_reports", # make sure this is the same as the index name you have used in SuperRag
10    }
11})
12
13agent = client.agent.create(request={
14    "name": "Financial Reports Agent",
15    "description": "An assistant that can answer questions about financial reports",
16    "isActive": True,
17    "avatar": "" # Optional avatar url jpg/png
18    "prompt": "Answer questions based on the financial reports",
19    "llmModel": "GPT_4_0613"
20})
21
22# Connect the tool to the agent
23client.agent.add_tool(agent_id=agent.data.id, tool_id=tool.data.id)
Built with