← Back
OpenAI
OpenAI and Amazon launch Stateful Runtime Environment for agents in Bedrock, simplifying production multi-step workflows
OpenAI APIOpenAI · featureapiintegrationplatform · openai.com ↗

Stateful Runtime Environment for Production Agents

OpenAI and Amazon have partnered to introduce a Stateful Runtime Environment that addresses a critical gap in agent deployment: moving from simple, stateless prototypes to production-grade, multi-step workflows.

What Changed

The new runtime environment is designed to run natively inside AWS environments and powered by OpenAI models. Unlike traditional stateless APIs that require development teams to manually orchestrate state, tool invocation, error handling, and task resumption, the Stateful Runtime Environment handles these concerns automatically.

Key capabilities include:

  • Persistent working context that carries forward memory, history, and workflow state across steps
  • Automatic tool invocation and error handling without manual scaffolding
  • AWS-native deployment with built-in security posture, governance rules, and tooling integrations
  • Identity and permission boundaries enforced at the runtime level
  • Long-running task reliability designed for complex, multi-step execution

What Developers Need to Know

This eliminates significant operational burden. Teams no longer need to build custom orchestration layers to manage:

  • Multi-step execution flow
  • State persistence between API calls
  • Tool/workflow state management
  • Long-running task resumption
  • Error handling and recovery

Use Cases Enabled

The runtime is tailored for production workflows including:

  • Multi-system customer support operations
  • Sales operations automation
  • Internal IT automation
  • Finance processes with approvals and audits

Getting Started

The Stateful Runtime is available soon in Amazon Bedrock. AWS customers can contact their OpenAI team or request more information to explore integration options.