Build composable, interoperable multi-agent systems with A2A on AWS
Machine Learning Blog
This article demonstrates building composable, interoperable multi-agent systems across organizational boundaries using the A2A (Agent-to-Agent) protocol on AWS.
- A2A protocol enables agents from different organizations and frameworks to discover, communicate, and collaborate without exposing internals or sharing secrets
- LangChain-based enterprise orchestrator communicates with Strands Agents-based SaaS compliance agent across separate AWS accounts
- Bidirectional OAuth 2.0 authentication using Amazon Cognito allows agents to act as peers, calling each other across trust boundaries
- Agent Cards enable runtime discovery of agent capabilities without requiring SDKs or documentation
- SaaS compliance agent calls back to enterprise context agent for metadata, demonstrating true peer-to-peer collaboration
- Deployment uses three Jupyter notebooks to provision infrastructure, deploy agents, and run end-to-end tests
- Security controls include transport encryption, guardrails, least-privilege IAM, container hardening, and account isolation
The solution shows how independent agents built on different frameworks can compose into workflows neither team built alone, enabling ISVs to replace custom integrations with a single protocol-based interface.
The AWS News Feed is currently looking for gold sponsors. If you want to support the AWS community and reach a large audience of AWS professionals, consider sponsoring the AWS News Feed.
Related articles
2026
2026
2026
2026
The AWS News Feed is currently looking for silver sponsors. If you want to support the AWS community and reach a large audience of AWS professionals, consider sponsoring the AWS News Feed.