AI Agents in Action

Author: Micheal Lanham Year: 2025 Genre: Artificial Intelligence

About This Book

AI Agents in Action teaches developers how to create and deploy agents and assistants built on large language models such as OpenAI’s GPT-4, Anthropic Claude, and Groq Labs Mixtral. You’ll learn to use prompt engineering to create agents with their own personas and profiles, set up multi-agent collaborations that can succeed in unpredictable environments, and build the kind of AI you can trust to handle high-stakes negotiations.

Key Insights

• Agent patterns over ad‑hoc prompts: Design task loops, memory, tools, and guardrails as first-class concerns.
• Tool use is the superpower: Reliable agents come from well-specified tools, schemas, and deterministic evaluators.
• Plans beat improvisation: Decompose work with planners, critics, and executors to improve reliability.
• Multi-agent when justified: Use collaboration patterns when they add capability; keep single-agent designs when simpler.
• Evaluation is engineering: Trace runs, compare configs, and iterate with offline evals before shipping.

Why I Recommend It

It’s a pragmatic bridge from “prompting” to robust agent systems. The book focuses on patterns, tooling, and evaluation you can apply immediately to ship dependable automations, not just demos.

If you’re building LLM-powered features, you’ll find concrete designs for tool calling, planning, memory, and testing that shorten the path to production.