Is PDF upload and correct translate to text possible?

Questions
, ,
1

:bullseye: What is your goal?

Goal is using PDF data translated to text in one shot analyze flow. Kept in one active scenario, and more or less cost free. Current scenario: Tally-OpenAi (vision) gatekeeper-OpenAi (create a com..), analyse module-Gmail

:thinking: What is the problem & what have you tried?

Vision-only chain — no PDF text extraction
Binary PDF = hard timeout, flow dies
No format-sorting modules (kept cheap/simple)
PDF needs OCR — different pipeline
Word (.docx) — same wall, zip-packed

3

Hey Marin,

sending the PDF for OCR and translating the result should work fine. Can you share some screenshots of the scenario you have and what errors you are getting?

4
  1. Tally-form receive both PDF and image-file. Gatekeeper redirect the PDF file to OCR and image-file url continues to OpenAi analyse module. The OpenAi module dont wait for OCR data to arrive, -the duality is the break point. (wait for data=time out)
  2. It is very important that numbers are matched exactly to the corresponding text. (eg. a paycheck is full of numbers referring to a tekst field in the Bottom of the doc. and so on)