ChatGPT for Extracting and Analyzing Payment Tables from PDFs/Images

Let’s see how this AI tool, which we can talk to, can extract and analyze a payment table (like a bank statement) from PDFs or images, focusing on non-USD transactions.

If PDFs don’t work, we’ll convert them to images.

So, here's what we'll explore:

• How to request data extraction by uploading a PDF;

• How to request data extraction by uploading an image or screenshot;

• How to request data extraction after converting a PDF to an image.

After extracting the data, we'll ask the chat to create an extended table with specific details.

So, we have a PDF file with just one page, showing daily payments to suppliers of a furniture business over two months. The file contains a table with the date, beneficiary, and amount:

________________________

We may:

• Upload the file directly into the chat;

• Copy and paste a screenshot of it into the same chat;

• Convert the PDF to an image (for example, requesting the PNG format).

Let's say we want to convert that PDF into a PNG:

________________________

Here's the answer:

________________________

We can open a new chat, upload the downloaded image, and request the extraction of the table. This approach can be useful when PDF processing is temporarily unavailable in the AI chatbot. Additionally, a screenshot or any other image format of the statement might work even better for this purpose.

In this instance, we requested the extraction within the same chat where we uploaded the PDF:

________________________

And here is the output, where ChatGPT provides us with a downloadable CSV file containing the extracted table:

________________________

Now, here’s the next request we can make to get a direct view in the chat:

________________________

And then we proceed with additional requests...

________________________

When we ask...

________________________

We get a downloadable file like this:

________________________

And we can request additional information to be added to another file, such as:

________________________

When we receive the updated file, here’s our final request:

________________________

The specific added section in the new file would appear like this:

________________________