Understanding and Building Agentic AI Systems: A Complete Beginner-to-Builder Textbook baixar epub

Everyone is talking about AI agents. Almost no one can explain how they actually work.

This book can — starting from zero. No degree, no programming experience, no prior knowledge of what an API, a database, or a server is: every term is defined the first time it appears, and one continuous story carries you from your first token to a production-grade AI agent.

You'll follow Maya, owner of a small online coffee roastery, and build Beans — the assistant that answers customers from her real documents with citations, looks up real orders, pauses subscriptions, and never refunds a dollar without a human's approval and a limit enforced in code.

• Part I — Foundations: what a model really does, why it makes things up, and the one sentence that explains every AI product you've ever used
• Part II — The modern stack, layer by layer: frontends, orchestration (LangGraph, CrewAI, Mastra), models, retrieval and vector databases (RAG), tools and MCP, the context builder, data, observability and evals, deployment, and security
• Part III — The complete step-by-step build: every command shown for both macOS and Windows, from installing Python to a running, tested agent — with checkpoints after every step and a troubleshooting table when something goes wrong
• Part IV — What endures: the ideas that outlive every logo, a production-readiness checklist, and exactly where to go next

• Honest labeling of durable principles versus product snapshots — so the book stays useful as tools change
• Safety as engineering: budgets, permission tiers, human-in-the-loop approvals, and prompt-injection defense enforced in code, not in promises
• The measurement habit: tracing and evaluations, so your system improves on purpose instead of quietly rotting
• A complete companion repository — every file of the Beans project, runnable and covered by 73 automated tests, matched to the book chapter by chapter
• An independent technical review of the entire build, published as an appendix, so you know exactly where a teaching demo ends and a production system begins

Career changers and students who want a real foundation, self-taught builders tired of fragmented tutorials, professionals whose teams are adopting AI, and anyone who has typed "how do AI agents work" and wanted a straight answer. Each chapter ends with a summary, key terms, and review questions — with a full answer key — so it works as a course text or a self-study guide.

By the last page you won't just use AI agents. You'll understand every layer of how they're built — and you'll have built one.