As most of you probably know, web imports are painfully slow at the moment. This is primarily due to the fact we need to spin up a web browser for each incoming request, open the page, wait for it to load, process it, THEN finally run our NLP analysis on it.

We'd toying with the idea of creating an importer browser extension that would allow you to import the webpage you're currently viewing, using the HTML already loaded in your browser. This would knock out the first three steps described above and theoretically make importing MUCH faster and easier.