Simple Chatbot agent
Running version [email protected]
running on n8n cloud
DB pinecone: metric cosine; dimensions: 1536
embedding_model:
text-embedding-3-small
The chatbot read Pinecone Vector store for answers. Everything functioned well until I added the functionality to put the chat messages into the vector store in a different name context.
Now it’s working, but each chat message comes with Metadata - for example:
Msg 1: Hello
Reply 1:
{ “metadata”: { “source”: “blob”, “blobType”: “application/json”, “loc”: { “lines”: { “from”: 1, “to”: 1 } } }, “pageContent”: “Hello! How can I assist you today?” }
The metadata disappears when I disable the vector store extension writing the sessions into the database.
Ideas?