RPA/screen scraping just uses the web site as the API. It’s a poorly defined, likely to change, poorly defined API. But it exists and accepts http.