N8n integration with Pinecone Assistant

[brat]

Building a chat bot and i have integrated AI Agent with Pinecone assistant . The assistant in Pinecone works with no issues , and gives me the response. However when i run the http request in n8n , it tells me it needs more context and returns nothing , I have verified the chat input and every possible detail. Not sure why the http integration is returning no results ..

[jennapederson]

Hi @brat - Pinecone Developer Advocate here! Can you share your workflow and the specific error message?

[brat]

Workflow is pretty simple. shown below and the error message . But the assistant itself in pinecone site works fine ..

[jennapederson]

Can you share the input to the node making that call? It would help to see that input format as well I suspect the input is malformed.

It looks like you’re using the chat API along with the AI Agent node which also does chat generation. This runs the risk of double the token usage and lower accuracy because assistant does chat generation and the ai agent node also does it. Instead I would suggest using either the context snippets API or the Assistants MCP server directly (via HTTP node) to get just context back to pass that to the AI agent’s model generation.

Here is a template that show how to do this: https://n8n.io/workflows/9942-rag-powered-document-chat-with-google-drive-openai-and-pinecone-assistant/

[Websensepro]

Hi @brat

This is a classic AI Agent setup problem. The agent responding “I need more context” is a key sign that it doesn’t know when to use your tool .

The Pinecone API call isn’t failing; the AI Agent is never even trying to run it. It decides whether to use a tool based on the tool’s Description . If the description isn’t clear, the agent will ignore the tool and try to answer on its own, which fails.

The Solution: Improve Your Tool’s Description

Open your “RagPodcastTranscripts” tool node (the one connected to the Tool input of the agent).
Find the Description field.
You need to write a very clear instruction here for the AI. Tell it exactly what this tool is for.

Bad Description: “Gets podcast data.”

Good Description: “Use this tool whenever a user asks a specific question about the content of a podcast. The input should be the user’s full question.”

Also, Check the Input Parameter:

In the same tool, under Input Schema , make sure you have defined a parameter for the user’s query. It should look something like this, which tells the agent how to pass the question to your HTTP Request:

[
  {
    "name": "query",
    "type": "string",
    "description": "The user's question about the podcast."
  }
]

By providing a clear description and a defined input, the AI Agent will understand exactly when to use your Pinecone tool and how to send the query. This will solve the “I need more context” error.

If you found this helpful, please mark it as the solution and give it a like .

[brat]

I integrated with MCP and it works fine now . Thank you .. Still not sure why the http request won’t work, i even tried @Websensepro suggestion .