I would like to transfer the functionality of a CustomGPT with a knowledge base consisting of PDF files in an n8n workflow. However, it doesn’t really work. I created an OpenAI node with ‘Message a model’. As far as I know, knowledge via PDF files cannot be included there. So I transferred the content of the PDF files to my Notion board and connected the content to the OpenAI node using tools. Results are coming out, but they are not half as good in quality as in CustomGPT, and in addition, not all tools are always used. I specifically added to the prompt that it should really run ALL tools before starting the task, but one or two of the seven tools are still missing. According to the documentation, the tools are not actually intended to replicate a knowledge base like CustomGPTs, so I fear that I may be on the wrong track. What is your experience? How can the functionality of a CustomGPT with PDF files as a knowledge base be effectively replicated in an n8n workflow?
To implement a CustomGPT simulation, you must use RAG, you can look to this official n8n workflow list about Retrieval Augmented Generation.
Hey @aipreneur,
I think there are lot of things that may have to be streamlined to get the same exact result as Custom GPT in n8n from prompt engineering to RAG retrieval. One workaround could be accessing the OpenAI response API node to get almost the same experience as Custom GPT in OpenAI. You can check it out here- https://platform.openai.com/docs/api-reference/responses
PS: If you need any professional help and a done for you service, I am Make Advanced Certified and you can contact me from my profile or through my website.
Cheers,
Sparsh
Founder
Automation Jinn
This topic was automatically closed 7 days after the last reply. New replies are no longer allowed.