AWS uses AI agents to cut cloud migration prep from weeks to minutes

Enterprise cloud migrations typically fail to meet timelines because of three structural bottlenecks that compound across large portfolios. Manual discovery—understanding on-premises architecture, inventory, and dependencies—consumes weeks per application. Writing infrastructure-as-code from scratch for each workload then takes another 3–4 weeks. And once migrated, teams fall into reactive operational mode, manually monitoring and responding to issues. Across over 300 applications on a fixed fiscal year deadline, these sequential delays become impossible to overcome with traditional approaches.

AWS Professional Services' multi-agent framework solves this by automating each phase while preserving human oversight. The Intake Agent ingests on-premises documentation and produces a target AWS architecture with migration patterns and sizing. That output flows directly—without manual handoff—to the IaC Agent, which generates infrastructure code adhering to security and compliance standards. The IaC generation step, where AWS deployed the framework first, delivered the most immediately measurable impact. A Migration Intelligence and Governance Agent then provides portfolio-wide visibility and compliance tracking, while an SRE Agent shifts post-migration operations from reactive to proactive with automated monitoring and remediation.

The architecture itself enforces security at every step: agents communicate through Model Context Protocol tools exposed via Amazon Bedrock AgentCore Gateway, authenticated by scoped IAM roles, and validated against Cedar policy rules before execution. Session state persists in AgentCore memory, allowing one agent's output to become the next agent's input seamlessly. This design turns what was years of engineering effort into an automated, auditable, and compliant process—the key to making aggressive migration timelines realistic for large enterprises.