Foundations of agentic AI on AWS
Aaron Sempf, Amazon Web Services
July 2025 (document history)
In a world of increasingly intelligent, distributed, and autonomous systems, the concept of an agent—an entity that can perceive its environment, reason about its state, and act with intent—has become foundational. Agents are not merely programs that execute instructions; they are goal-oriented, context-aware entities that make decisions on behalf of users, systems, or organizations. Their emergence reflects a shift in how you build and think about software: a shift from procedural logic and reactive automation to systems that operate with autonomy and purpose.
At the intersection of AI, distributed systems, and software engineering lies a powerful paradigm known as agentic AI. This new generation of intelligent systems consists of software agents that are capable of adaptive behavior, complex coordination, and delegated decision-making.
This guide introduces the principles that define modern software agents and outlines their evolution toward agentic AI. To explain this shift, the guide provides the conceptual background and then traces the evolution of software agents to agentic AI:
-
Introduction to software agents defines software agents, compares them to traditional software components, and introduces the essential characteristics that differentiate agentic behavior from traditional automation by drawing from established frameworks.
-
The purpose of software agents examines why software agents exist, what roles they fulfill, what problems they solve, and how they enable intelligent delegation, reduce cognitive load, and support adaptive behavior in dynamic environments.
-
The evolution of software agents traces the intellectual and technological milestones that shaped software agents, from early concepts of autonomy and concurrency to the emergence of multi-agent systems and formal agent architectures, resulting in the convergence with generative AI.
-
Software agents to agentic AI introduces agentic AI as the culmination of decades of progress that combines distributed agent models with foundation models, serverless compute, and orchestration protocols. This section describes how this convergence enables a new generation of intelligent, tool-using agents that operate with autonomy, asynchronicity, and true agency at scale.
Intended audience
This guide is designed for architects, developers, and technology leaders who want to understand the history, main concepts, and evolution of software agents to agentic AI before they adopt this technology for modern cloud solutions on AWS.
Objectives
Adopting agentic architectures helps organizations:
-
Accelerate time to value: Automate and scale knowledge work, and reduce manual effort and latency.
-
Improve customer engagement: Deliver intelligent assistants across domains.
-
Reduce operational costs: Automate decision flows that previously required human input or oversight.
-
Drive innovation and differentiation: Build intelligent products that adapt, learn, and compete in real time.
-
Modernize legacy workflows: Reframe scripts and monoliths into modular reasoning agents.
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 series includes the following:
For more information about this content series, see Agentic AI
