AWS’s Swami Sivasubramanian is building the cloud for AI agents

When Swami Sivasubramanian joined Amazon as a technical intern in 2005, cloud computing was just getting off the ground as a concept. Thanks to Amazon Web Services, it grew into one of technology’s most essential ingredients—and Sivasubramanian, who kept rising through the ranks, helped drive that change. In 2017, he became AWS’s VP of AI. More recently, as large language models permitted artificial intelligence to perform ever more complex tasks with greater independence, he had an epiphany: “AI agents will fundamentally change how we all work and live,” he says. To accelerate AWS’s agentic efforts, he spearheaded the formation of a new group focused entirely on the technology. Like other AWS offerings, the company’s new agent services let customers offload deployment and pay as they go for computing that scales with demand. Last year, it announced Amazon Bedrock AgentCore, an agent-building platform already used by Ericsson and Thomson Reuters to automate manufacturing. Another new service, Kiro, lets DevOps teams create agents to analyze and manage infrastructure. Many organizations are still just figuring out which tasks to hand off to agentic AI. Sivasubramanian notes that one common use case—modernizing legacy mainframe software—involves more than the obvious task of translating the COBOL language into Java. It also means untangling poorly documented systems made up of countless modules. That leaves AWS needing a clear view not only of the things companies know they want to achieve with AI, but also the surprises they may encounter along the way. “Nine out of 10 times, we do exactly what customers want,” he says. “And one out of 10 times, we read between the lines and [conclude] they’re asking for a faster horse instead of a car. Then we build a car.” This profile is part of Fast Company’s AI 20 for 2026, our roundup spotlighting 20 of AI’s most influential technologists, entrepreneurs, corporate leaders, and creative thinkers.