Intelligent computer implemented agents are associated with computer user interface tasks by dividing the tasks into statistically distinct clusters based on sampled user assessments. The assessments collect data on multiple user variables. Multivariate statistical analysis is used to divide the tasks into distinct clusters. The clusters are validated using univariate analysis on each of the measured variables. Intelligent agents are associated based on the measured variables to ensure that agents are effective. The objective assessment and association avoids costly creation and overhead of agents applied where not effective.