ChatGPT-4 with website but without webscraping

I want to know if there is a way to connect chatgpt-4 with attendees to a website, which is what I want to do, that when a customer uses a chatbot they can ask questions and take the answers from a website.

I already did it with webscrapping by downloading the html of the site and uploading the files to the assistants but I would like to skip this step.

Do you have an idea how to achieve this?

How do you recon the model could get access to your data if you don’t want to give it your data? :thinking:

if you have a popular website and didn’t update the data since the last training cutoff, it might work out of the box. but other than that, how could it possibly work?

Did you mean that you do not want to scrape the page and add it as file?

There is another way. If you know the page to check, you can just invoke a function call that will web scrape it and send you the info back. This way, you can make your bot adapt to whatever website you want. However, you need to handle if the site has very big text data. You can also probably scrape, chunk it in the background, get embedding and do RAG. The question is, how will your bot know what website? You can probably set the site URL in the additional_instructions parameter when you run a thread.

Yes, in this case a field is enabled in the settings where the website URL is injected, but I want it to search directly on the website, not search in files.