MATSIM-T consists of several (logically seperated parts). All modules are based on MATSIM-DB which contains all available data points needed for modeling agents and their demand.
World-DB, Network-DB, Facilities-DB, etc... describes the scenario and aggregated information like OD-matrices and so on. MATSIM-FUSION fuse those diverse data into a well deformed data sturture stored into memory.
Based on the scenario information MATSIM-INI them models each individual in the scenario including individual initial demand, called plans (i.e. for a day). The Agents are stored into Agent-DB which refers to the given scenario information of the other databases of MATSIM-DB.
In MATSIM-EA the individual demand of each agent will be optimized using an iterative process. The process steps consists of:

• MATSIM-EXEC: It is the physical agent-based mobility-simulation in which the agents actually interacts with others
• MATSIM-SCORES: These modules measure the performance of an executed demand of each agent
• MATSIM-STRATEGY: Strategy modules changes/mutates/generates executed demand of agents according to the information of the last execution of the demand.

By iterating these steps the overall performance of the system (sum of scores of each plan of each agent) increase until a stable state is reached.
The result are optimized demand of each agent and - as a side effect - the exact description of the last execution of this individual demand. Therefore, at every point in time, we know where and agent is and what it is doing at that time.
These very precice information can then be aggregated via MATSIM-ANALYSIS to extract desired information like traffic volumes, departure histograms, system relaxations, location occupation, and so on.