You determine at the outset how much of your time you want to spend engaged with the interface, and how you prioritize different types of activities (I want to spend 10% of my time reading articles, 10% reading books, 20% following up with people, 25% coding, etc). You then engage the interface daily for the period of time you determined, and the interface directs your actions to maximize your priorities.

As a user, you would still be able to move freely between different types of tasks, but the interface would promote the highest priority activity that matches the types of activities the user is in the mood for.

This would be a sort of dance between the user and the interface. Sometimes the interface and the user are in exact agreement, the highest priority activities are also the ones you haven’t made progress on recently and are the activities you are gravitating towards anyways. Sometimes the user simply isn’t in the mood to do the work that the interface is suggesting is highest priority, and the user can opt for a different task, albeit with a bit more effort than if the user were to do the highest priority work.

This tweaking of the required effort to take certain actions is the key to the system. The interface should make the most important things to do the easiest things to do, making our action as effortless as possible. This is the key to what I call System Design, and what Devon Zuegel calls Incentive design.

The interface’s prioritization would use as much information as possible. Perhaps the interface thinks I should do my article reading, as I’ve been falling behind on that for the last week, and I’m not spending the desired 10% of time I said I wanted to spend on reading articles. I also am in the mood for book reading, so when I see it presented to me I happily accept.

Much like other smart-assistant features, this type of interface would use all the data it has at its disposal to make the best decisions it can about how to guide the user.