Autonomous learning agents need careful design and tooling to achieve useful levels of interaction. This lecture connects autonomous systems with recent ideas around agentic models. These autonomous systems are key to achieving the goal of learning from real-world interactions (reinforcement learning).