دانلود کتاب مقدس سیستم‌های هوش مصنوعی عامل‌محور – ساخت عامل‌های هوشمند با پایتون

This book was born at the convergence of two technological revolutions: the maturation of Large Language Models (LLMs) and the emergence of a new paradigm in software engineering — autonomous agentic systems (Agentic AI Systems). The first revolution is by now well known. From GPT-3 (Brown et al., 2020) to modern reasoning models such as o3 and Claude Opus, LLMs have transformed the way we interact with computational systems. These models, trained on billions of parameters with self-attention mechanisms (Vaswani et al., 2017), have acquired capabilities that were considered theoretically impossible just a few years ago: natural language understanding, multi-step reasoning, code generation, and complex data analysis. The second revolution is less visible but arguably more significant. It is the transition from passive question-answering models to autonomous systems that perceive, reason, and act — the classic Perceive-Reason-Act loop described by Russell and Norvig (2020) in their foundational textbook. An LLM by itself is a passive mechanism: it receives input, produces text, and terminates. An AI agent, by contrast, maintains state, uses tools, makes decisions, and iterates — until it completes its objective. This distinction is not academic. It is the difference between a chatbot that tells you how to send an email and an agent that sends it itself. Between a model that analyzes data if you provide it, and a system that searches for the data, processes it, and delivers a report. Between a tool that answers questions and a system that solves problems. This field is experiencing explosive growth. Companies like OpenAI, Anthropic, Google DeepMind, and Microsoft are investing billions in developing agent frameworks and infrastructure. Industry is beginning to adopt agentic workflows in critical applications: automated customer service, document analysis, IT infrastructure management, software engineering, and scientific research. According to analyst estimates, the AI agents market is expected to exceed 65 billion dollars by 2030. Despite this, knowledge about how to build these systems remains fragmented. There are academic papers, blog posts, tutorials, code examples — but no comprehensive work that guides the reader from the foundations to production-ready systems. This is precisely the gap this book aspires to fill.