Agentic AI patterns and workflows on AWS - AWS Prescriptive Guidance
Intended audienceObjectivesAbout this content series

Agentic AI patterns and workflows on AWS

Aaron Sempf and Andrew Hooker, Amazon Web Services

July 2025 (document history)

Organizations are adopting large language models (LLMs) and software agents to solve dynamic, multidomain problems using a new architectural discipline called agentic patterns. Agentic patterns are foundational blueprints and modular constructs that are used to design and orchestrate goal-oriented AI agents across many contexts.

Intended audience

This guide is intended for architects, developers, and product leaders who want to build intelligent applications that go beyond static logic, symbolic logic, and deterministic automation.

Objectives

This guide provides a design framework and implementation approach for AI agent systems that operate autonomously while remaining controllable and aligned with your goals. It connects event-driven architectural patterns with various agentic alternatives, demonstrating how to build production-grade agent systems using cloud-native architectures. The following subjects are discussed in this guide:

  • Agent patterns – Agent patterns are reusable design templates that describe the structure and behavior of individual agents. This includes reasoning agents, retrieval-augmented agents, coding agents, voice interfaces, workflow orchestrators, and collaborative multi-agent systems. Each pattern illustrates how agents perceive, reason, act, and learn, mapped to AWS services.

  • LLM workflows – Workflows focus on how agents use LLMs for reasoning. They explore prompting strategies and planning mechanisms, and outline how LLMs are used not only to generate text but also to drive structured, interpretable, and reliable behaviors within an agent loop.

  • Agentic workflow patterns – Workflow patterns describe how multiple agents, tools, and environments interact to form autonomous systems. This includes patterns for task orchestration, subagent delegation, event-based coordination, observability, and control. These aspects promote scalable, composable, and auditable AI architectures.

About this content series

This guide is part of a set of publications that provide architectural blueprints and technical guidance for building AI-driven software agents on AWS. The AWS Prescriptive Guidance series includes the following guides:

For more information about this content series, see Agentic AI.