From what I’ve found the past few hours working on this, I hit a limitation. I’m trying to pass a binary file from a parent workflow to a sub-workflow. The file comes through, but when I try to pass it to my AI agent, it’s not working.
The idea is:
OpenAI’s latest model n8n AI Agent node should accept multi-modal binary file payloads.
My use case:
I need to pass a file (pdf, docx or image) to my AI agent as, or included in, my prompt.
I think it would be beneficial to add this because:
Everyone would love this.
Any resources to support this?
GPT-4o can now directly accept files (PDF, DOCX, images) with the /v1/chat/completions endpoint, using tool_choice: auto.
Are you willing to work on this?
If i got paid to.