Building AI Agents

From Design Patterns to Production
Erscheinungsjahr 2027
ISBN: 978-3-032-39095-0
Verlag: Springer

Buch, Englisch, Format (B × H): 155 mm x 235 mm

Most AI agent demos work flawlessly, until real users arrive. The gap between a compelling prototype and a dependable system is where this book begins. Framing agent development as an engineering discipline, it centers on the core loop of Perceive, Plan, Act, and Observe as the foundation for building agents that are robust, interpretable, and scalable.

As a companion to Antonio Gulli’s , carries the field’s emerging architectural vocabulary from concept into practice. Where the earlier volume defines the patterns, this book implements them: showing how to translate agent design principles into reliable, production-ready systems.

Organized around four proven architectural patterns -- ReAct, Chain-of-Thought, Reflection, and Plan-and-Execute -- the book takes a hands-on, pattern-first approach. Each chapter includes working code and contributes to the development of Atlas, a unified research and coding assistant that evolves from a minimal script into a production-grade multi-agent system.

Coverage spans tool integration, memory and state management, multi-agent orchestration, and system evaluation. Implementations are demonstrated across leading frameworks, including LangGraph, CrewAI, OpenAI’s Agents SDK, and Google’s ADK, with designs that generalize across major model providers such as OpenAI, Gemini, Claude, and Llama. The emphasis throughout is on transferable patterns rather than vendor-specific solutions.

This book is intended for software engineers integrating agents into production systems, AI/ML practitioners moving beyond chat-based interfaces, technical leads evaluating architectural tradeoffs, and advanced students working at the frontier of applied AI. Readers should be comfortable with Python and have a foundational understanding of large language models; all other concepts are developed in context.

Zielgruppe

Professional/practitioner

Autoren/Hrsg.

Gulli, Antonio

Nawalgaria, Anant

Fachgebiete

• Mathematik | Informatik EDV | Informatik Informatik Künstliche Intelligenz
• Mathematik | Informatik EDV | Informatik Programmierung | Softwareentwicklung Programmier- und Skriptsprachen

Inhaltsverzeichnis

Chapter 1: Anatomy of an Agent.- Chapter 2: Prompt Architecture for Agents.- Chapter 3: Tools, Skills, and Structured Outputs.- Chapter 4: Handoffs and Routines.- Chapter 5: Stateful Agent Graphs.- Chapter 6: Multi-Agent Collaboration.- Chapter 7: Model Portability: One Agent, Many Models.- Chapter 8: Open Protocols: MCP, A2A A2UI, and AP2.- Chapter 9: Declarative Agent Skills.- Chapter 10: Building Agents with Claude Code and Google Antigravity.- Chapter 11: Memory and Agentic RAG.- Chapter 12: Code Execution and Sandbox Agents.- Chapter 13: Multimodal and Voice Agents.- Chapter 14: Guardrails and Agent Safety.- Chapter 15: Agent Harness Engineering.- Chapter 16: Always-On Agents: Daemon Loops, Watchdogs, and Graceful Recovery.- Chapter 17: Managed Agents: Let the Platform Run It.- Chapter 18: Evaluation and Observability.- Chapter 19: Deployment, Async Agents, and Security.- Chapter 20: Loop Engineering: The System Is the Loop.- Chapter 21: Capstone: Atlas — The Autonomous Engineering Assistant.- Chapter 22: What's Next: Agentic AI in 2026 & Beyond.- Appendix A.- Appendix B.

Über Autor(innen)

Antonio Gulli is a highly experienced Senior Director at Google, currently leading the Engineer Director role in the Office of CTO. With over 30 years of relevant experience, Antonio is a well-known figure in the industry, with a strong background in AI, Search, and Cloud technologies. Antonio has extensive experience managing technical teams and providing Google Cloud technology solutions across EMEA industry. He has previously served as Site Lead and Engineering Director for Google, where he managed cloud teams and led cross-functional teams in strong collaboration with international sites and sales functions. Antonio has also authored the book "Deep Learning for Keras" to increase science culture awareness. Antonio's educational background is impressive, with a Ph.D. in Computer Science from the University of Pisa and a Master's degree in Engineering from the same university. He also holds a Master's degree in Practice Engineering from the University of Pisa and a Bachelor's degree in Computer Science from the University of Pisa. Antonio's technical expertise includes Senior Software Engineer, AI, Search, Cloud Kubernetes, Keras, and Deep Learning. He is also a Board Member and VC Advisor, making him a valuable asset to any organization. 

Anant Nawalgaria is a Senior Staff Machine Learning Architect and Product Leader at Google. Specializing in bridging the gap between cutting-edge AI research and scalable production, his work spans generative AI, multi-agent systems, and novel architectures for multimodal LLMs. Anant has co-authored research spotlighted at prestigious conferences such as ICLR and ICML.

At Google, Anant drives 0-to-1 research and product creation, leading pioneering technologies like Gecko, a state-of-the-art multimodal evaluation framework, and AlphaEvolve, an advanced evolutionary coding agent utilizing black-box optimization. He has also contributed to core enhancements of early Gemini models, researching novel architectures for performance and cost optimization in multimodal LLMs. Beyond research, Anant leverages his forward-deployment engineering expertise to guide some of the world's largest, most innovative enterprises in building and scaling cutting-edge AI solutions for novel production use cases.

Anant is also a prominent technical educator and industry voice. He founded and led the global Google x Kaggle Generative AI course series, reaching over two million learners worldwide, has spoken at numerous industry-wide events, and publishes highly popular technical blogs and white papers on generative AI, agents, and vibe coding. Additionally, he serves as a contributor and reviewer for multiple authoritative books on applied machine learning and agentic systems. Anant holds a Master of Science in Informatics (with Distinction) from the Technical University of Munich (TUM) and an MBA from the Indian Institute of Technology Madras (IIT-M).