Hi Intercom community,
We're running into a limitation with Fin that I'd love to get some input on.
Fin currently only processes attachments shared as images (screenshots/photos) — it can't read the contents of document files like PDFs, DOCX or xlsx uploaded by customers via email or conversations.
This is a significant blocker for our invoice workflow. Here's our setup:
- We're training an AI Billing agent in Langdock, with Fin handling triage by extracting the relevant details from customer messages.
- We've built out internal articles describing each document type, plus an index covering invoice formats, use cases, field structures, etc.
- In most cases, customers simply mention their invoice number in the message — and that works fine.
- However, when a customer uploads a PDF instead, Fin still asks for the invoice number, even though that information is right there in the attachment.
Ideally, Fin would be able to read the PDF, extract the invoice number (and other relevant fields), and pass that context to the billing agent — without putting the burden back on the customer.
Has anyone found a workaround for this? Whether it's a middleware solution, a Zap/webhook approach, or something else entirely — we're open to ideas. Any help is appreciated!
Best regards
Matej
