Agentic Design Patterns: A Hands-On Guide to Building Intelligent Systems
by Antonio Gullí (Springer Nature)
21 agent design patterns, from prompt chaining to A2A, each with runnable LangChain, CrewAI and Google ADK code
Overview
Agentic Design Patterns was published by Springer Nature on October 31, 2025 (XLIII + 427 pages; $39.99 eBook on Springer's page as of October 2026) and is written by Antonio Gullí, a Senior Director and Distinguished Engineer in Google's Office of the CTO with over 30 years in AI, search and cloud and a PhD in computer science from the University of Pisa. The book has four parts plus appendices. Part One covers core execution patterns: prompt chaining, routing, parallelization, reflection, tool use, planning and multi-agent collaboration. Part Two moves to memory management, learning and adaptation, the Model Context Protocol, and goal setting and monitoring. Part Three handles production concerns with exception handling and recovery, human-in-the-loop and knowledge retrieval (RAG). Part Four adds inter-agent communication over A2A, resource-aware optimization, reasoning techniques, guardrails and safety patterns, evaluation and monitoring, prioritization, and exploration and discovery. Seven appendices cover advanced prompting, GUI and real-world agent environments, an overview of agentic frameworks, building an agent with AgentSpace, agents on the CLI, reasoning engines and coding agents. Every chapter follows the same template: a pattern overview, practical applications, one or more code examples, and key takeaways. The code targets LangChain and LangGraph, CrewAI and Google's Agent Development Kit, so you can compare how each framework expresses the same pattern. On Goodreads (3.83 average from 78 ratings, checked October 6, 2026), positive reviews value it as a single reference that collects the patterns, while critical reviews call it repetitive and too shallow for experienced engineers.
At a Glance
- Topic
- Agentic
- Level
- Intermediate
- Format
- Book
- Cost
- Paid
- Duration
- 427-page book (21 pattern chapters + 7 appendices), self-paced
- Provider
- Antonio Gullí (Springer Nature)
- Hands-on
- Yes — code/exercises
- Certificate
- None
What You’ll Learn
- ✓Chain prompts and route requests between specialized LLM calls based on intent
- ✓Run subtasks in parallel and add reflection loops that critique and revise outputs
- ✓Implement tool use (function calling) and planning patterns for multi-step agent tasks
- ✓Design multi-agent collaboration and inter-agent communication using the A2A protocol
- ✓Add memory management to agents and connect them to external tools over MCP
- ✓Handle exceptions, recovery and human-in-the-loop checkpoints in production agent workflows
- ✓Apply guardrails, safety patterns, evaluation and monitoring to agents you deploy
- ✓Compare how LangChain/LangGraph, CrewAI and Google ADK implement the same pattern
Highlights
- •One chapter per pattern with a fixed template (overview, use cases, code, key takeaways), so it works as a lookup reference
- •Code examples span three frameworks, LangChain/LangGraph, CrewAI and Google ADK, so you can see one pattern expressed three ways
- •Covers protocol-level patterns such as MCP and A2A alongside prompt chaining, reflection and planning
- •Written by a Google Distinguished Engineer in the Office of the CTO; Goodreads reviewers are split, with critics citing repetition and thin analysis for senior engineers
Who It’s For
Best For
- ✓Developers moving from single LLM calls to multi-step agents
- ✓Teams that want a shared vocabulary for agent architecture reviews
- ✓Engineers comparing LangGraph, CrewAI and Google ADK
Prerequisites
- •Working knowledge of Python
- •Basic experience calling an LLM API and writing prompts
FAQ
What is Agentic Design Patterns: A Hands-On Guide to Building Intelligent Systems?
A 427-page Springer book by Google's Antonio Gullí that catalogs 21 design patterns for LLM agents, one chapter per pattern, each with use cases and runnable code. It suits developers who already call LLM APIs and want a shared vocabulary for routing, planning, memory, tool use, guardrails and multi-agent setups before they commit to a framework.
Is Agentic Design Patterns: A Hands-On Guide to Building Intelligent Systems free?
Agentic Design Patterns: A Hands-On Guide to Building Intelligent Systems is a paid resource.
What level is Agentic Design Patterns: A Hands-On Guide to Building Intelligent Systems for?
Agentic Design Patterns: A Hands-On Guide to Building Intelligent Systems is aimed at a intermediate audience. Recommended background: Working knowledge of Python, Basic experience calling an LLM API and writing prompts.
How long does Agentic Design Patterns: A Hands-On Guide to Building Intelligent Systems take?
Expect roughly 427-page book (21 pattern chapters + 7 appendices), self-paced. Most learners work through it at their own pace.
What will I learn from Agentic Design Patterns: A Hands-On Guide to Building Intelligent Systems?
You'll learn: Chain prompts and route requests between specialized LLM calls based on intent; Run subtasks in parallel and add reflection loops that critique and revise outputs; Implement tool use (function calling) and planning patterns for multi-step agent tasks; Design multi-agent collaboration and inter-agent communication using the A2A protocol; Add memory management to agents and connect them to external tools over MCP; Handle exceptions, recovery and human-in-the-loop checkpoints in production agent workflows; Apply guardrails, safety patterns, evaluation and monitoring to agents you deploy; Compare how LangChain/LangGraph, CrewAI and Google ADK implement the same pattern.
Topics
Sources
This page was written from 5 sources, 4 on domains other than link.springer.com.