We used Pocket to build our training set, but now we need a streaming feed of articles to run our model against. To set this up, we'll use IFTTT once again, as well as Google Sheets, and a Python library that will allow us to work with Google Sheets.