Ask HN: Need Help with Data Pipeline

5 pointsmrbishalsaha2 comments

Hey guys, I am building a data pipeline to securely extract data from bank statements, credit reports and other financial docs to feed it into a schema for my database so we can work on that data.

Currently I see ChatGPT has the best OCR, but I can't make it to perfectly. Can anyone help. What should I do?

View on HN

> I am building a data pipeline to securely extract data from bank statements

> Currently I see ChatGPT has the best OCR

> What should I do?

Think hard about the combination of “securely” and using ChatGPT, in either meaning of the word. If you want to keep the data private, you don’t want to send it out on the internet anywhere. If you want your OCR to perform repeatable, you don’t want to use ChatGPT in its current form. It will every now and then make up reasonably convincing lies.

When done, think about whether HN is the proper place to ask these questions.

Where should I then ask such questions?

Reply on news.ycombinator.com