Introducing the Stateful Runtime Environment for Agents in A…
OpenAI NewsAI 代理在推理方面表现出色，但更棘手的是落地执行：如何在真实工具和系统上，长期可靠地运行多步骤任务，并建立恰当的管控机制。
今天，我们通过与 Amazon 的合作，推出可原生运行于 Amazon Bedrock 、由 OpenAI 提供动力的新型 Stateful Runtime Environment ，以简化这一过程。 AWS 客户将能通过该 Runtime Environment 获取 OpenAI 提供的智能，这一环境针对 AWS 基础设施 进行了优化，并为 agentic workflows 量身定制，提供生产级所需的状态管理、可靠性和治理能力。
简化代理的生产化流程
许多基于无状态 API 的代理原型只应对简单场景：一次提示、一次回答，或可能一次工具调用。而生产环境不同：真实的工作流往往由多步组成，需要沿用先前操作的上下文，依赖多个工具输出、审批和系统状态，并且必须在安全环境中设置可信的防护措施。
无状态 API 要求开发团队自行构建支持性编排层。团队必须解决状态如何存储、工具如何调用、错误如何处理、以及长期运行任务如何安全恢复等问题，这增加了开发负担。
为减轻这类负担， Stateful Runtime Environment 经由双方联合设计，可在你的 AWS environment 内运行，并针对 AWS services 进行了优化。现在，代理不再需要开发者手工拼接孤立请求；它们能够自动执行复杂步骤，携带向前传递的“ working context ”——包括记忆/历史、工具与工作流状态、环境使用情况，以及身份/权限边界。
这些能力带来的可能性
部署多系统的客户支持、销售运营工作流、内部 IT 自动化，以及需要审批与审计的财务流程，比以往更容易实现。
加速多步骤工作流的生产化
当运行时负责跨步骤的持久编排与状态管理后，团队可以把精力放在工作流和业务逻辑上，而不是搭建支撑框架。
更适合长期、跨时段的任务
有状态的任务被设计为能随时间可靠运行，持续带动执行所需的上下文与控制边界。
原生 AWS 部署与治理
该运行时旨在在客户的 AWS environment 内运行，便于遵循既有的安全策略、工具集成与治理规则。
入手方式
Amazon Bedrock 中的 Stateful Runtime 很快就会上线。若想了解它将如何助力你的 AI 之旅，请联系你的 OpenAI 团队 或提交联系请求。
名（必填）
姓（必填）
电子邮件（必填）
电话号码
公司名称
公司规模（必填） 请选择一个选项 1–50 51–250 251–500 501–1,000 1,001–5,000 5,001–10,000 10,001–20,000 20,001+
你能否分享更多有关业务需求与挑战的信息？（最多 500 字符）
我愿意通过电子邮件接收来自 OpenAI 的产品、服务和活动相关的营销信息。我理解可随时通过相关营销邮件中的退订链接或访问 go.openai.com/Unsubscribe.html 撤回同意。
提交
AI agents excel at reasoning. The harder part is operational: running multi-step work reliably over time, across real tools and real systems, with the right controls.
Today, we’re making this easier for customers through a partnership and joint collaboration with Amazon to deliver the new Stateful Runtime Environment that runs natively in Amazon Bedrock, powered by OpenAI. AWS customers will have access to OpenAI intelligence delivered through the Runtime Environment, optimized for AWS infrastructure and tailored for agentic workflows, with the state, reliability, and governance needed for production work.
Simplify getting agents into production
A lot of agent prototypes based on stateless APIs tackle simple use cases: one prompt, one answer, maybe one tool call. Production work is different. Real workflows unfold across many steps, require context from previous actions, depend on multiple tool outputs, approvals, and system state, and need trusted guardrails in secure environments.
Stateless APIs require building the supporting orchestration layer for agents. The burden is on development teams to figure out how state is stored, how tools are invoked, how errors are handled, and how long-running tasks resume safely.
The Stateful Runtime Environment is jointly designed to reduce that load. It runs inside your AWS environment and is optimized to work with AWS services. Now, instead of manually stitching together disconnected requests to make things work, your agents automatically execute complex steps with “working context” that carries forward memory/history, tool and workflow state, environment use, and identity/permission boundaries.
What this unlocks
It’s now easier than ever to deliver solutions like multi-system customer support, sales operations workflows, internal IT automation, and finance processes with approvals and audits.
Faster time to production for multi-step workflows
When the runtime handles persistent orchestration and state across steps, teams can focus on the workflow and the business logic instead of scaffolding.
A better fit for long-horizon work
Stateful tasks are designed to run reliably over time, carrying forward the context and control boundaries needed for multi-step execution.
AWS-native deployment and governance
The runtime is designed to operate within customers’ AWS environment to easily comply with the existing security posture, tooling integrations, and governance rules.
Getting started
The Stateful Runtime in Amazon Bedrock will be available soon. If you want to explore how this will benefit your AI journey, reach out to your OpenAI team or request to be contacted.
First name *
Last name *
Email *
Phone number *
Company name *
Company size *
Select an option1-5051-250251-500501-1,0001,001-5,0005,001-10,00010,001-20,00020,001+
Can you share more about your business needs and challenges?
Maximum 500 characters
I would like to receive marketing communications from OpenAI via email about its products, services and events. I understand that I may withdraw this consent at any time by clicking the unsubscribe link in a related marketing communication or by visiting go.openai.com/Unsubscribe.html
Submit
Generated by RSStT. The copyright belongs to the original author.