This is a great question! I think the hipster thing to do is say AI/ML but I actually have an Operations Research background and I can tell you the word AI is overloaded.

I have some practical thoughts and I'd love to see if you folks think it will work.

The big idea is to leverage three pieces of information:

1. The due date the tasker agrees to do, 2. The date in which the task actually gets done, 3. Resets of the date one way or another.

In general, the system continuously catalogues the difference between 1 and 2, and 1 and 3 and computes an average +/-. The +/- is then shown to anyone who attempts to request a due date on the individual. People who are well loaded would have +/- close to 0 and those who are overloaded would have more negative numbers. My thought is that just exposing these numbers would allow people to factor in more reasonable asks for the individual much like showing one's CPU load.

These are good points - may not work for more creative/specialized teams where the tasks are all relatively unique and not well quantified.

Might work for teams that have layers of people in the same role (e.g. Supervisors) all doing more or less the same tasks (e.g. interviewing new hires). Things like manufacturing, IT ops, sales ops, service industries...