Agentic AI Engineering : Build secure and cost-effective autonomous systems with agent harness and production-ready patterns

Agentic AI Engineering : Build secure and cost-effective autonomous systems with agent harness and production-ready patterns

  • ただいまウェブストアではご注文を受け付けておりません。 ⇒古書を探す
  • 製本 Paperback:紙装版/ペーパーバック版
  • 言語 ENG
  • 商品コード 9781807604714
  • DDC分類 006.3

Full Description

Build reliable AI agents with the OpenAI Agents SDK using Harness, Graph Engineering and Loop Engineering patterns with working examples, and production controls for evaluation, and security

Key Features

Build single-agent and multi-agent systems with reusable patterns
Apply evaluation, security, tracing, and cost controls from day one
Integrate enterprise tools and agents using MCP and A2A
Purchase of the print or Kindle book includes a free PDF eBook

Book DescriptionMove beyond impressive demos and build agentic AI systems that remain reliable with real users, sensitive data, compliance requirements, and budgets. This book gives you durable architectural patterns, working code, and production controls for creating AI agents that can be tested, secured, monitored, and governed.

You'll build single-agent and multi-agent systems with the OpenAI Agents SDK using planner/executor, router, supervisor/worker, memory-tier, tool-registry, and human-in-the-loop patterns. You'll connect agents to tools and enterprise services using function calling, MCP, and A2A, while addressing side effects, access control, shared state, model routing, and failure recovery.

The book treats evaluation, observability, security, and cost as core engineering disciplines. You'll create evaluation harnesses and regression gates, apply loop engineering principles, defend against prompt injection and memory poisoning, trace agent decisions, control token spend, and deploy with versioned prompts, gradual rollouts, and multi-tenant isolation.

Written by Mohsen Amiribesheli, PhD, an AI consultancy director who has delivered production AI solutions for Fortune 500 enterprises, this guide shares lessons from regulated industries. By the end, you'll be able to design, deploy, and operate trustworthy agentic systems at scale.What you will learn

Design agents with durable architectural patterns
Build agents using the OpenAI Agents SDK
Connect tools and services through MCP
Coordinate multi-agent workflows with A2A
Test agent behavior with evaluation harnesses
Apply guardrails and least-privilege controls
Trace performance, failures, and operating costs
Deploy and evolve reliable agent systems

Who this book is forThis book is for AI engineers, application developers, solution architects, technical leads, and engineering managers who need to take AI agents beyond the prototype stage and into production. You should be comfortable reading Python and know your way around APIs, prompting, and tool calling. No prior agent framework experience is required.

Contents

Table of Contents

Anatomy of an Agent: How Agentic Systems Are Really Built
From Zero to Agent: Building with the OpenAI Agents SDK
Giving Agents Hands: Tools, Function Calling, and MCP
What Agents Remember: Memory, Knowledge, and Context
Who Runs the Show: Orchestration Patterns and Model Routing
Agents That Collaborate: Multi-Agent Systems and A2A
Does It Actually Work? Evaluating and Testing Agent Systems
Keeping Agents in Line: Security, Guardrails, and Governance
Watching the Money: Observability, Cost, and the ROI Case
Shipping It: Deployment Patterns and Infrastructure
Keeping It Alive: Operating and Evolving Agent Systems

最近チェックした商品