Agents can utilize tools to access real time data and to invoke business logic. Agentic applications can even generate execution plans how multiple agents should be orchestrated for specific tasks. While the capabilities are evolving quickly, agents don't always work reliably. Developers and AI Engineers need to understand how these systems work to improve them. Because of the level of complexity, observability for agents is even more important than for Large Language Models only. In this session you'll learn why observability and transparency are key topics for agents and how the open-source agent framework 'Bee' helps developers to understand agentic flows.