Books

AI agents are only as good as the loops behind them. This book shows you how to build systems that plan, act, evaluate and improve with every step. Learn the patterns, frameworks and engineering practices behind reliable, self-correcting agents that solve real problems in production.

Build powerful AI agents that run entirely on your own hardware with Ollama. Learn practical patterns for RAG, multi-agent systems, security and deployment through clear explanations and production-ready code you can use right away. From first setup to reliable real-world applications, this book helps you build with confidence.

Every news feed is full of AI buzzwords: Transformers, tokens, embeddings, context windows, hallucinations, objective functions. Yet most explanations are either dense academic textbooks or empty marketing fluff. ☕ Coffee Break AI is your practical guide to AI. Written in plain English with warm real-world analogies. It breaks down the core mechanisms of AI into 40 bite-sized chapters.

Et si vous pouviez construire un logiciel avec Codex sans être ingénieur ? Ce livre explique comment transformer une idée en mission claire, comment ChatGPT prépare les prompts transmis à Codex et comment garder le contrôle grâce aux tests, à Git et à des preuves simples.

Building an impressive AI agent demo is easy. Running one in production—reliably, at scale, with governance—is where most enterprises get stuck. Agentic Orchestration shows how BPMN-based process orchestration governs AI agents at scale, drawing on real deployments across banking, telecom, and healthcare. Reasoning without orchestration is expensive chaos. This book shows the way out.

Die Nutzung von KI zum Programmieren ist einfach. Sie zu nutzen, ohne die Kontrolle zu verlieren, schon weniger. Spec-Driven Development ist die Methode, um deine Idee in eine Spezifikation zu verwandeln, die die KI präzise ausführt — ohne Endlosschleifen, ohne kaputten Code, ohne wieder bei null anzufangen.

16 章手写一个 CLI Agent。从 Hello World 到 Skill 插件系统，只用 Python 标准库 + OpenAI SDK。每章附真实项目踩坑笔记——来自 git 提交历史的 bug、设计失误和凌晨调试的教训。读完你会理解 Agent 的底层运作方式，而非只是会用某个框架。

The future of AI depends on how we measure it. This book cuts through the hype to show what benchmarks really tell us, where they fall short and why better evaluation leads to better models. Practical, clear and grounded in real-world experience, it is an essential guide for anyone building, studying or deploying modern AI.

A £4.1M PRA compliance failure. A fictional bank called Avon & Wessex. Sixteen chapters and 23 systems showing exactly how to build production-grade, auditable AI for credit, market, operational and liquidity risk, model governance, and regulatory reporting — the practitioner's guide UK banking has been missing.

Evaluation Engineering for AI Systems is a practical guide to building reliable AI evaluations for production. Learn how to measure performance, compare models, detect regressions and create evaluation pipelines using real-world examples, modern Python and proven industry practices.

What if TransE, ComplEx, RotatE and the rest of the knowledge graph “model zoo” were different views of one geometric operator? Learn the mathematics, code and practical design principles behind structured memory for trustworthy AI.

Discover how to unlock the full potential of Claude Opus 5 with a practical guide built on Anthropic's official documentation and proven best practices. From writing your first effective prompt to designing sophisticated agentic workflows and production-ready systems this book gives you the techniques, patterns and insights needed to achieve more accurate, reliable and consistent results.

An agent can score well on average and still fail exactly where it matters. Beyond “Ship and Pray” shows how to replace benchmark averages with designed experiments, geometric ground truth and failure attribution—so teams can discover when an agent breaks, identify the responsible component and test whether it fails safely under tool faults.

Top-k is not relevance, retrieved text is not evidence and an LLM judging another LLM is not verification. Beyond “Chunk and Pray” shows how to build RAG that answers through a verified knowledge graph, preserves exact numbers, cites its sources and abstains when it cannot prove the answer.

Build AI agents that can see screens, reason about tasks and control desktop applications and web browsers. Starting with a simple prototype, you'll learn how to create reliable production-ready systems that do real work on a computer.