内容来源：https://aibusiness.com/agentic-ai/prompt-ai-agents-becoming-operational-infrastructure

内容总结：

谷歌云特约报道：生成式AI进入企业“实操阶段”，安全与治理成焦点

随着人工智能从概念演示走向企业日常运营，各大组织正面临因AI系统自主性提升而带来的治理、基础设施与运维新挑战。专家指出，切入生成式AI的第一步，应聚焦于能够改善人类信息获取体验的领域。

本周，AI智能体正加速嵌入真实工作流：亚马逊云服务（AWS）为自主智能体新增支付功能，使其能直接完成交易；Salesforce旗下Tableau推出智能分析平台，传统商业智能工具正向上下文感知的AI转型；IBM宣布企业AI进入新阶段，从孤立应用转向跨业务场景的协同部署；领英将求职搜索升级为AI语义检索，支持自然语言对话；Ace Hardware则为门店员工推出AI助手，实时提供产品信息与项目指导。

然而，随着智能体从实验走向生产，企业也趋于谨慎。许多组织虽对AI充满期待，却对伴随而来的安全与治理挑战准备不足。智能体的持续性、协同性与日益增强的自主性，正在倒逼企业重构底层基础设施，并围绕AI系统重新设计工作流程。

与此同时，Anthropic本周与SpaceX达成算力协议，以扩充Claude Code及API的调用容量。这些动向共同揭示了一个趋势：AI的下一个阶段，不再是仅响应指令的系统，而是能主动采取行动、贯穿业务流、甚至成为企业运营一部分的“行动者”。因此，核心议题正从“能力”转向“控制”——如何治理、保护并管理这些越来越自主的系统，将成为定义企业AI未来走向的关键。

本周AI领域其他要闻：

• 英伟达借助机器人生态系统推动“物理AI”规模化，布局真实世界自主系统。
• 思科与施耐德电气呼吁完善监管政策，以支持制造业等行业的AI负责任落地。
• 亚马逊推出商品摘要交互功能，将对话式AI更深融入零售体验。
• 科技行业裁员潮引发关于AI如何快速改变劳动力需求的广泛讨论。
• IBM扩展面向混合云与大型机的企业AI智能体策略，推进业务级AI协同部署。

中文翻译：

赞助商：谷歌云
选择您的首个生成式AI应用场景
要开始使用生成式AI，首先应聚焦于那些能改善人类与信息互动的领域。

随着AI代理从演示阶段进入企业工作流，各组织正面临由更自主的AI系统引发的治理、基础设施和运营挑战。

编者按：欢迎来到《Prompt》，这是为您提供的每周AI领域动态简报。我们既会对本周重大进展进行深度分析，也会精选真正重要的资讯。

AI的下一个阶段不仅关乎模型，更在于代理走向实际运营。
AI代理正从实验阶段进入企业基础设施和运营环节，各组织正竭力思考如何对其进行治理、保障安全并实现运营化。

代理正走向真实的工作岗位，深度融入工作流程而非停留在演示层面。仅本周，我们就看到了多个实例：

• AWS为自主代理新增支付功能，使AI系统能在企业工作流中完成交易并采取更直接的操作。
• Salesforce旗下的Tableau推出其智能代理分析平台，体现了传统BI厂商如何借助AI代理和情境感知分析实现进化。
• IBM表示企业AI正进入新阶段，从独立应用转向更为协调、可运营的跨业务AI代理部署。
• LinkedIn将其求职体验转向AI驱动的语义搜索，让用户能以更接近自然对话的方式进行搜索。
• Ace Hardware为门店员工推出AI助手，让工作人员能实时获取产品信息、项目指导和推荐建议。

这一转变正使企业在权衡风险与回报时变得更加谨慎——许多组织渴望使用AI，却未准备好应对随之而来的安全性和治理挑战。

基础设施层正因代理而改变。代理因其持续性、协调性和日益增强的自主性，给基础设施带来了复杂性。这些变化促使企业围绕AI系统重新设计工作方式。

与此同时，Anthropic本周与SpaceX达成的计算资源协议，使其生成式AI实验室得以扩展计算能力，并立即提升了Claude Code和Claude API的使用上限。

综合来看，这些发展指向了企业内部AI使用方式的更广泛转变。
AI的下一个阶段并非仅仅是回应提示语的系统，而是能够采取行动、跨工作流运行并日益成为业务本身的系统。

这使得讨论焦点从能力转向控制。随着代理更深地嵌入运营，挑战已不再是构建更强大的模型，而是如何治理、保障并管理那些日益具有更高自主性的系统。

这将是定义企业AI下一阶段的关键所在。

本周AI领域其他动态：
除代理之外，报道还聚焦AI如何开始重塑零售、劳动力战略、企业基础设施及真实世界中的自主系统。

• Nvidia依托机器人生态扩展物理AI：Nvidia正借助其更广泛的机器人生态系统助力扩展物理AI，突显业界对真实世界自主系统的兴趣日益浓厚。
• 思科与施耐德电气呼吁制定有利法规以促进AI发展：在SelectUSA投资峰会上，发言者表示，虽然AI有潜力重塑制造业等行业，但需要更有力的政策来支持负责任的应用并建立技术信任。
• 亚马逊最新AI功能允许购物者与产品摘要互动：亚马逊的新AI购物功能反映了对话式AI如何更深入地融入零售体验。
• 科技行业裁员显示AI替代已在发生：新一轮科技行业裁员正引发讨论：AI正在以多快的速度改变整个行业的劳动力需求。
• IBM通过代理推动混合云与大型机上的企业AI：IBM正扩展其面向代理、混合云和大型机的企业AI战略，推动在商业环境中实现更协调的AI部署。

英文来源：

Sponsored by Google Cloud
Choosing Your First Generative AI Use Cases
To get started with generative AI, first focus on areas that can improve human experiences with information.
As agents move past demos and into enterprise workflows, organizations are confronting the governance, infrastructure and operational problems posed by more autonomous AI systems.
Editor’s Note: Welcome to Prompt, your weekly briefing on the shifting AI landscape. We provide an analytical look at the week’s biggest developments, paired with a curated roundup of the stories that actually matter.
The next phase of AI isn’t just about models. It’s about agents becoming operational.
AI agents are moving from experimentation into enterprise infrastructure and operations, and organizations are now trying to figure out how to govern, secure and operationalize them.
Agents are moving into real working roles, becoming embedded in workflows rather than remaining limited to demos. This week alone, we’ve seen several examples of that:
AWS added new payment capabilities for autonomous agents, enabling AI systems to complete transactions and take more direct action inside enterprise workflows.
Salesforce-owned Tableau introduced its Agentic Analytics Platform, reflecting how traditional BI vendors are evolving with AI agents and context-aware analytics.
IBM said enterprise AI is entering a new phase, moving beyond isolated applications toward more orchestrated, operational deployment of AI agents across the business.
LinkedIn shifted its job search experience toward AI-driven semantic search, letting users search more conversationally using natural language.
Ace Hardware launched an AI assistant for store employees, giving staff real-time access to product information, project guidance and recommendations.
This shift is causing enterprises to become more cautious as they try tobalance risk and reward, with many organizations eager to use AI butunprepared for the security and governance challenges it entails.
The infrastructure layer is changing because of agents. Agents are adding infrastructure complexity because they’re persistent, orchestrated and increasingly autonomous. Those changes have companies redesigning work around AI systems.
Meanwhile, Anthropic’s compute deal with SpaceX this week enables the generative AI lab to expand compute capacity, immediately expanding usage limits for Claude Code and the Claude API.
Taken together, these developments point to a broader shift in how AI is being used inside the enterprise.
The next phase of AI isn’t about systems that simply respond to prompts. It’s about systems that can take action, operate across workflows and increasingly function as part of the business itself.
That changes the conversation from capability to control. As agents become more embedded in operations, the challenge is no longer just building more powerful models. It’s now a matter of figuring out how to govern, secure and manage systems that are increasingly acting with greater autonomy.
That’s where the next phase of enterprise AI will be defined.
Also in AI This Week:
Beyond agents, coverage highlighted how AI is beginning to reshape retail, workforce strategy, enterprise infrastructure and real-world autonomous systems.
Nvidia Taps Robotics Ecosystem to Scale Physical AI: Nvidia is leaning on its broader robotics ecosystem to help scale physical AI, highlighting growing interest in real-world autonomous systems.
Cisco, Schneider Electric Call for Enabling Regulations to Help AI Flourish: While artificial intelligence has the potential to reshape industries such as manufacturing, speakers at the SelectUSA Investment Summit said stronger policies will be needed to support responsible adoption and build trust in the technology.
Amazon’s Latest AI Feature Allows Shoppers to Interact With Product Summaries: Amazon’s new AI shopping feature reflects how conversational AI is becoming more embedded in the retail experience.
Tech Sector Job Losses Show AI Replacement in Action: New rounds of tech-sector job cuts are fueling debate over how quickly AI is starting to alter workforce needs across the industry.
IBM Pursues Enterprise AI With Agents for Hybrid Cloud, Mainframes: IBM is expanding its enterprise AI strategy for agents, hybrid cloud and mainframes as it pushes for more orchestrated AI deployment across business environments.