Thank you! We are collecting many data sources from the whole web and the backend concurrently fetches the status. We use SSE to return the data, just like ChatGPT. And the speed could potentially be improved by around 30%.