I want to use ChatGPT to access a website that provides SOAP API documentation from an ERP system. It should help me to retrieve data from the ERP via the API and visualize it, for example. However, ChatGPT always reports that it cannot access this page because it contains JavaScript and the browser must support frames (which, for example, the Edge browser actually supports). I have tried the GPT-4 with Bing search, but also plugins like Link Reader, Webpilot etc. or GPTs like Scraper, Web Scraper etc… None of these methods can read the website. This is also the case with many other websites.
Does anyone know of a solution that makes ChatGPT crawl/scrape websites that require JavaScript and frames to serve data for different use cases? Or are there intermediate solutions that can provide ChatGPT with such data?
I would be very happy to hear about possible solutions.
Of course, I don’t want to enter the data manually. ChatGPT should fetch the data/parameters for the database, which are not small, itself.
And as I said, not a single browsing plugin or scraping GPT has worked so far.
Correct. I have also give the LLM the URL for the frameless version of the SOAP-API Documentation. But this also not working.
I want the LLM to read the documentation and produce code or give me the parameters that I need to request data from the database for analytics, bi etc.
You will need to build custom functions to access the api. What you can do is give GPT the API docs and have it help you write the code. I did that for our system.
There might also already be library (SDK) to access your system via API out there.
Then you would need to use these as function calling (tools) inside a custom script or somehow expose “actions” to a custom GPT. a lot of steps so without some programming background it will be hard