Automating Enterprise Cloud Migrations

AWS Professional Services has highlighted how organizations can scale large cloud migrations using agentic artificial intelligence. According to the AWS Machine Learning Blog, a multi-agent framework built on Amazon Bedrock AgentCore helps automate enterprise cloud migrations end to end across complex portfolios.

Large-scale enterprise migrations often face strict fiscal year deadlines and require custom automation alongside managed services. By deploying purpose-built AI agents to handle discovery, infrastructure as code generation, portfolio governance, and post-migration operations, development time for infrastructure as code can be drastically reduced from several weeks down to mere minutes.

Hybrid Architecture and Core Services

The migration framework operates alongside established AWS offerings rather than replacing them. AWS Transform covers migration and modernization work for server, network, mainframe, .NET, and application code workloads, while AWS Database Migration Service handles the database tier.

To handle organization-specific requirements, the custom agents attach to these managed services. Sources and destinations are reached through Model Context Protocol tools that are built and maintained by the enterprise delivery team. Developers build these purpose-built agents using the Strands Agents SDK and run them on Amazon Bedrock AgentCore, a platform designed to build, connect, and optimize agents at scale.

The Four-Agent Pattern and Workflow

The multi-agent architecture incorporates four specialized agents mapped across the migration and operations journeys. The Intake Agent reads migration inputs from document and collaboration systems using Model Context Protocol tools to define the target state architecture during the initial phase.

Following the intake phase, the IaC Agent generates infrastructure code that composes approved internal modules for each application. Meanwhile, the Migration Intelligence and Governance Agent reports and governs inside the organization's own program tools, and the Site Reliability Engineering Agent handles operations after cutover.

Runtime Environment and Security Controls

Amazon Bedrock AgentCore runtime hosts the Strands agents in a serverless environment equipped with session isolation and multi-agent orchestration. Foundation models available through Amazon Bedrock power the reasoning required to interpret documents, generate code, and drive multi-step workflows.

Each agent calls Model Context Protocol tools scoped to its specific function through AgentCore Gateway, which converts APIs and existing services into compatible tools. Furthermore, AgentCore Identity authenticates each call using scoped AWS Identity and Access Management roles, while Amazon Bedrock Guardrails applies responsible AI controls to the agent outputs.

Sources

• AWS Machine Learning BlogScaling cloud migrations with agentic AI on Amazon Bedrock AgentCore