"AWS Bedrock AgentCore: Memory, Tools, and Event-Driven Agents on AWS"
Built for senior cloud engineers, platform architects, and experienced AI practitioners, this book is a deep technical guide to designing production-grade agent systems with Amazon Bedrock AgentCore. Rather than stopping at prompts or simple chat workflows, it focuses on the hard parts of real deployments: long-running execution, durable memory, secure tool use, event-driven orchestration, and the operational realities of agents that act across systems over time.
Readers will learn how AgentCore Runtime, Memory, Gateway, MCP, and AWS-native event services fit together into a cohesive architecture. The book explains how to choose AgentCore over simpler integrations, engineer resumable workloads, design short- and long-term memory boundaries, expose tools safely, and build reactive workflows driven by runtime, tool, and memory events. It also covers identity, delegated permissions, multi-tenant isolation, observability, evaluation pipelines, and the feedback loops needed to continuously improve agent behavior in production.
The treatment is architectural and implementation-oriented, with strong emphasis on trade-offs, trust boundaries, and operational decision-making. Familiarity with AWS, distributed systems, APIs, IAM, and modern LLM application patterns will help readers get the most from it. For teams moving from experiments to durable platforms, this book provides the vocabulary, design framework, and system-level discipline required to build reliable agents on AWS.
