OpenAI and Amazon today unveiled a new Stateful Runtime Environment running natively in Amazon Bedrock, designed to simplify deploying production-grade AI agents. This collaboration addresses a critical gap in current AI toolchains: while stateless APIs work for simple tasks, real-world workflows require persistent context across multiple steps, tool integrations, and secure governance. The new runtime handles this orchestration automatically, eliminating the need for custom scaffolding and reducing development overhead for complex agent deployments. AWS-native deployment ensures seamless integration with existing infrastructure, allowing customers to leverage their current security and compliance frameworks while building multi-system workflows for customer support, IT automation, and financial approvals.

By managing state, error recovery, and tool execution across steps, the runtime lets teams focus on business logic rather than infrastructure plumbing. This is particularly valuable for long-horizon tasks where maintaining context and permissions across sessions is essential. The solution operates within customers' AWS environments, maintaining their existing security posture while providing the reliability needed for production use. Early access is available for customers exploring how this improves their AI journey.

This partnership represents a significant step toward making advanced agent capabilities more accessible for enterprise applications, directly tackling the operational challenges that have limited the adoption of multi-step AI systems in real-world scenarios.