@Diego (WMF) Please review outputs 1 and 2 under Outcome 1 and: link them to their corresponding phabricator tasks if they're not already linked, and also let me know if we have missed a goal altogether, and whether the current ones I listed capture what we discussed.
One thing I noticed for Output 1 is that I'm not sure if we discussed to identify both the types of feedback we should collect from the users for improvement /and/ setting up the model to be able to react to the feedback. We definitely did talk about the latter. If we want to do the former, which in retrospect makes sense to me, we should check with Jonathan if he has time to help. @Jmorgan (WMF) FYI.