For some reason the RAG agent which is meant to trigger every minute is not triggering every minute. it is activated. When I run the insert to supabase node module manually to trigger the workflow, its running through the entire workflow and updating the database with extracted text from PDF, but the chat function is saying it cant answer my question, even though I can clearly see it has retrieved a relevant chunk from the vectorstore. I’ve almost given up on this workflow. any help appreciated!
Which model are you using? I’ve seen this happen a lot with smaller ones (<8b parameters).
Hi Cole. Im using 4o mini so thought that should be robust enough and is strange as the vector store is retrieving the correct chunks. But the other issue is that its not checking my google drive every minute for updates or new files, how do I fix that?
thank you!
Yeah 4o mini should be enough! Just in case though I would try with regular 4o and see if that changes anything. I don’t think RAG in n8n is optimized the best tbh.
Is your workflow toggled as active? You can see this in the top right. It has to be active to be watching your Google Drive folders.
yep was active. I dropped a new file in the folder and waited, but wasnt triggering at all. Played around with different settings for half an hour and no luck. But strangely, I tested again just now and it worked! Possibly a connection issue with railway.app?
Yeah it could be! That’s super weird. I feel like I’ve had this happen with n8n before where it didn’t work for a while and then all the sudden did… wish I knew what caused it. But I’m glad it is working for you now
1 Like