This document proposes using an ant colony optimization (ACO) algorithm to automatically generate test cases for graphical user interfaces (GUIs). GUIs are becoming more complex, making manual GUI testing labor intensive. Traditional scripting approaches have limitations. The approach models GUI testing as an optimization problem, using ACO to search for test sequences that maximize coverage of the call tree within the system under test. This allows generating test sequences online during execution rather than requiring an explicit GUI model. The framework executes sequences of actions on the GUI to evaluate coverage without risking infeasible sequences.