Summary
In this chapter, we learned about the crucial role of tools and planning in AI agent systems. We discussed what tool/function calling is and how LLM agents exhibit this property. We also learned about various tool types and saw examples of how to use tools with frameworks or natively with an LLM. Subsequently, we explored various planning algorithms, from traditional approaches such as STRIPS and HTN to modern LLM-based planning methods, understanding their relative practicality in the context of language models. Through a practical travel planning example, we saw how tools can be defined, integrated, and utilized within each framework to create sophisticated planning systems.
We learned how integrating tool calling with planning can supercharge agentic systems by making them more capable of handling complex tasks. We also reviewed the implementation patterns across three frameworks (CrewAI, AutoGen, and LangGraph), which revealed distinct approaches to agent coordination...
