New Book Details 'Agentic Architectural Patterns'
A new book on "Agentic Architectural Patterns" was just released, aimed at enterprise AI architects. The book reportedly covers the GenAI landscape, patterns for multi-agent coordination, the spectrum from RAG to fine-tuning, and robust design patterns for implementing LLMOps at scale.
The book is authored by Dr. Ali Arsanjani, Director of Applied AI Engineering at Google Cloud, and Juan Pablo Bustos, a technology leader with a background at Google, Stripe, and Amazon Web Services. Arsanjani is a recognized pioneer in the field, often called the "father of SOA" (Service-Oriented Architecture), and holds foundational patents in service decomposition and context-aware routing. A core concept introduced is the "GenAI Maturity Model," a framework for organizations to assess their readiness and map a strategic path toward adopting agentic AI. The book details a progression from foundational systems to production-ready services and ultimately to self-improving AI ecosystems. The text moves beyond single-agent systems to detail hierarchical multi-agent architectures. It introduces design patterns where high-level "orchestrator agents" manage complex business workflows by delegating tasks to specialized agents. For inter-agent collaboration, the book proposes a specific "Agent-to-Agent (A2A) protocol" to ensure that autonomous systems can communicate and remain transparent, auditable, and grounded in ethical governance. This directly addresses the challenge of managing and scaling complex AI interactions within an enterprise setting. The architectural stack described includes three key layers: function calling, tool protocols like Multimodal Composite Platforms (MCPs), and the A2A protocol for agent collaboration. This provides a structured approach for building robust and interoperable AI systems. Practical application is emphasized through real-world case studies and code examples that utilize open-source frameworks. The book is designed to be accessible to professionals with a basic grasp of data and software concepts, aiming to bridge the gap from experimental prototypes to sophisticated, autonomous enterprise systems.
