Early AI Agent research primarily focused on Multi-Agent Systems and reinforcement learning domains, such as robot navigation and game theory scenarios.