I have wanted for a very looooong time the ultimate productivity tool. Okay, mostly I've thought of it as the ultimate TODO/scheduler/calendar application, but as I was thinking about it the other day, it's even more than that.

I'll track the evolution of my thoughts around it here. Hopefully in the near future, I'll see if I can't get it built.

Here's my initial brainstorm about it:

It is Task List with buckets, Calendar, rule-based Scheduler, mind map, project management, reminder, life process reminder/helper/urger/encourager, alarm clock, memory retention strategies, personal assistant who learns your idiosyncrasies, incorporate some NLP concepts/techniques as appropriate, etc.

I like the idea of things all get dumped into one bucket initially, though the interface should make it trivial to immediately send it to another bucket.

It would also be really nice if for the initial entry, any extra context that can be auto gleaned would be extremely nice. It's easy to type in 4 words into the initial bucket and then later forget what it was about. But… we want to minimize keystrokes and even more minimize clicks and navigation.

Tasks, concepts, etc., would default to go into the initial bucket but trivially can be moved to any bucket…

hmm… the whole bucket system could be a mind map, you could even move and alter the buckets like a mind map - and that fractally. So mind map of buckets of mind maps of buckets, etc.

Moving things around perhaps should be handled by orthogonal axes - chronological, importance, urgency, size, type, etc.

Should be some nice guides for decomposition of tasks - and from a conceptual level, not just implementation - something there… be creative.

At some point, individual tasks can have effort estimated. Could use either calendar time, or some pseudo unit that the software learns the conversion for. In fact the user could say - it's twice as hard as this other task, and that's enough. Also, should incorporate concept of energy - highly difficult tasks or ones prone to frustration should take more energy and thus be clustered with low energy tasks. Also, they will be the ones higher risky and more prone to inaccurate estimates.

The user could input and the software could learn how when he works best on certain types of tasks to make the rule based scheduler extremely effective and useful. For example, Maybe a user always fails to do certain tasks in the afternoon - the scheduler will try to schedule those types of tasks in the morning.

Support win/grateful/want/target/etc. lists.

Integrate email, IM, text messaging, etc.