Extracting Data from ChatGPT API Without Python – Alternatives for SAP Integration?

API
1

Hi everyone,

I’m using the ChatGPT API from an SAP environment and need to extract data from various document formats, including PDF, Word, and Excel. However, I’d prefer not to use Python in this setup.

I understand that the text completion endpoint can process text-based inputs, which might work for OCR-extracted image text. However, as far as I know, it does not natively support direct file processing for formats like PDF, Word, or Excel.

Given that constraint, are there any alternative methods, tools, or APIs (possibly within the OpenAI ecosystem or third-party solutions) that could help me extract structured data from these files before sending it to the ChatGPT API?

Any guidance, best practices, or workarounds for handling this within an SAP-based environment would be greatly appreciated!

Thanks in advance!

2

Do you mean the chatcompletions API ? If that’s the case, then there is no ability to upload files via that api. You have a few options :

  1. Using the chatcompletions API - Leave your unstructured data outside and process using techniques such as RAG using an external vector database and provide that as context to the chatcompletions API
  2. Use Assistants API for your use case and use the built in vector database provided by OpenAI.
  3. I always recommend trying your usecase using Playground or Custom GPT ( depending on your usecase)
1 Like
3

Actually you can send images to it. Just not pdf, docx or xlsx…

You need a document processing pipeline not just a RAG.

1 Like