# AI Agents in Action by [[Micheal Lanham]] ## Summary <!-- a couple of paragraphs --> Micheal Lanham's *AI Agents in Action* is a hands-on guide to building LLM-powered agents — systems that reason, call tools, plan, and act autonomously toward a goal. It moves from the basics of connecting a model to external functions through to multi-step reasoning, memory, and orchestration of multiple agents. The book treats the agent as a loop: the model observes, decides, acts via tools, and incorporates results, iterating until the task is complete. A distinctive thread is the progression of reasoning strategies — from basic prompting up through chain-of-thought, self-consistency, and tree-of-thought search — and how each increases an agent's ability to tackle harder problems at the cost of more compute. Lanham also addresses the practical pitfalls: agents that loop, hallucinate tool calls, or compound small errors into large ones. The result is a pragmatic engineering view of what it takes to make agents reliable enough to use. ## Table of Contents - Introduction to agents and their world - Harnessing the power of large language models - Engaging GPT assistants - Exploring multi-agent systems - Empowering agents with actions (tool/function calling) - Building autonomous assistants - Agent reasoning and evaluation (chain-of-thought, self-consistency, tree of thought) - Agent planning and feedback - Mastering agent prompts and memory - Agent memory and retrieval ## Notes <!-- main takeaways; LINK to the permanent notes this book grounds --> - Grounds [[Tool Use]] and [[Function Calling]] — how an agent invokes external capabilities. - Supports [[Tool Taxonomy]] (the kinds of tools agents are equipped with). - Backs [[Agent Planning]] and [[Reflection]] as mechanisms for multi-step autonomy. - Grounds the reasoning ladder: [[Chain-of-Thought]] → [[Self-Consistency]] → [[Tree of Thought]]. - Underpins [[Agent Failure Modes]] (looping, hallucinated tool calls, [[Compounding Error]]). ## Quotes - <!-- placeholder: add a verified short quote here --> ## Relevance to the course - Primary grounding for Module 2 — agent architecture, tool use, planning, and the progression of reasoning strategies. --- ## References -