Agent based AI

We can assume that this will be a relatively knowledge-rich system, so its content should (approximately) parallel (some) of the PlanetMath content -- but we will need more structure to make simulations work.

To define: what is an agent?

What sorts of reasoning steps and heuristics can they use? Etc.

Note: some inspiration can come from the Peeragogy project's design pattern catalog! http://peeragogy.org/practice -- under the assumption that mathematics is made up of lots and lots of projects at all levels