AI Agent model output empty

Describe the problem/error/question

I’m trying to use an AI Agent node to query a Pinecone vector store. The Run 1 through the Model keeps returning no output I can see in the chat panel or when checking on the AI Agent node.

What is the error message (if any)?

None

Please share your workflow

Share the output returned by the last node

Output appears to fail to search the vector store correctly and always returns a sorry we couldn’t find it style response.

Information on your n8n setup

  • n8n version: 1.95.3
  • Database (default: SQLite): SQLite
  • n8n EXECUTIONS_PROCESS setting (default: own, main):
  • Running n8n via (Docker, npm, n8n cloud, desktop app): npm
  • Operating system: Ubuntu

To ensure your data is properly embedded and indexed:

The Pinecone Vector Store node in “Insert Documents” mode sends the text to the index. Then, using “Get Many” mode, you can run manual search tests. If it works here, your embeddings and data are properly indexed. The “Populate a Pinecone vector database from a website” example in the documentation shows how to scrape, embed, and query without agents.

The documentation specifies that the Pinecone Vector Store node should connect directly to the AI ​​Agent’s tools connector using the “Retrieve Documents (As Tool for AI Agent)” mode. In that mode, you can define the name and description, which are visible to the model and tell it when to use the tool.

Before integrating it into the agent, connect the Pinecone Vector Store node to the Retrieve Documents node and run a test query (Prompt + number of results). If you get relevant documents, the vector store is working properly.

This tutorial may also help you: