The document describes KohonAnts, an algorithm that combines concepts from Kohonen's Self-Organizing Maps and Ant Colony Optimization to perform clustering and pattern classification. It works by associating input samples to "ants" that move on a grid, updating pheromone values to cluster similar samples together. Experiments on iris, glass and diabetes datasets show it outperforms KNN and other methods, with promising results for clustering and classification tasks. Future work includes additional comparisons to other algorithms and testing on more complex problems.