Colony is an educational resource designed to explain the attention mechanism of Large Language Models through simple analogies involving agents. The simulation places these agents within a board environment inspired by Conway's Game of Life. Each agent in the system represents a specific role within the self-attention block mechanism of an LLM. This visual approach allows users to observe how information flows and interacts during the attention process. The project is available as an open-source tool for those interested in exploring these concepts without complex mathematics. It serves as a fun and accessible way to understand the internal workings of transformer models.