AI-powered agents are seen as the ultimate goal of AI, leveraging foundation models to perform tasks autonomously. An agent perceives its environment and acts upon it using available tools. The success of an agent depends on its access to tools and the strength of its AI planner. Agents can use tools for knowledge augmentation, capability extension, and environmental action. Planning is a crucial aspect, and can be coupled with execution. Evaluation is about detecting failures in planning, tool execution, and efficiency, with the goal of creating more reliable AI systems.