Advice for a pdf information extraction tool

Need help
1

Hi All,

expert advice needed on how to best implement this feature.

Feature:
User upload financial statement for company in pdf for analysis.
Action:
Extract financial data from the financial statement like sales, cogs, expenses, profit etc for analysis

What is the best way to implement this ?

Two options I am thinking are:

Option 1:

  • Google Vision API to extract the text
  • Send text to AI like chatGPT or Claude to process and return the information back to Bubble.

Option 2:

  • implement a chat to PDF type stack using AI or use an existing API provided by some of the services available.

What would be the best approach taking the following into consideration.

  • Speed
  • Accuracy
  • Privacy

Thanks
Ty

2

Go to huggingface, there you can find lots of models which are specifically trained to extract data from financial reports you can use them