内容来源：https://aibusiness.com/agentic-ai/aws-bets-frontier-agents-as-next-era-enterprise-ai

内容总结：

【云科技伦敦报道】在亚马逊云科技（AWS）伦敦峰会上，该公司正式将“前沿智能体”（Frontier Agents）定位为下一代企业AI的核心方向。这类自主运行、可长期执行任务的系统，被视为超越传统辅助工具的颠覆性能力。

AWS专业服务与智能体AI副总裁弗朗西斯卡·瓦斯奎兹在主题演讲中指出，前沿智能体具备三大核心特征：自主性、大规模扩展能力和持久运行能力。“你可以为它们设定目标，它们会自主规划实现路径，同时处理多项任务，甚至连续工作数小时或数天。”

作为这一战略的落地产品，AWS去年推出的智能体开发平台Kiro已实现通过自然语言指令自主编写代码。该公司表示，此举旨在解决软件开发工具规模化过程中“能力与可控性脱节”的痛点。AWS同时展示了DevOps智能体与安全智能体，可在软件构建过程中自动诊断错误并扫描漏洞。

英国二手车交易平台Motorway成为Kiro的典型应用案例。该平台首席工程师瑞安·科马克透露，目前超过80%的工程师每日使用Kiro，每月生成代码量超百万行。“我们不是用AI颠覆工作方式，而是用它加速完成我们想做的事。”为确保代码质量，Motorway建立了标准化审核流程，要求工程师在规划阶段严格把控Kiro的代码生成方向。

在环境监测领域，AWS与伦敦自然历史博物馆合作部署传感器网络，实时采集城市生态数据。该项目已积累约800万个数据点，用于分析高温、交通等因素对生物多样性的影响。AWS欧洲、中东及非洲区市场负责人希拉里·谭表示，该“首个实景实验室”旨在将环境数据转化为政策制定者和企业的可执行洞察，推动可持续发展从成本中心向创新业务模型转型。

中文翻译：

由谷歌云赞助
选择您的首个生成式AI用例
要开始使用生成式AI，首先应关注能够改善人类与信息交互体验的领域。
这家科技巨头正将具备自主性的长期运行智能体定位为企业AI的下一个定义性变革。
伦敦讯——亚马逊云科技已牢牢抓住"前沿智能体"这一概念，这类系统将人工智能从单纯的辅助工具提升为能够完全自主完成复杂任务的智能体。
在本周亚马逊云科技伦敦峰会的主旨演讲中，该公司专业服务与智能体AI副总裁弗朗西斯卡·瓦斯奎兹围绕三大核心能力阐述了这些系统：自主性、规模性和持久性。
"前沿智能体是能力显著更强的新型智能体，"瓦斯奎兹表示，"你可以为它们设定目标，它们会自行规划实现路径。它们具备大规模并行处理能力，能同时执行多项任务，并可连续工作数小时甚至数天，以追求那些宏大且有时目标模糊的任务。"
在此背景下，亚马逊云科技去年推出了Kiro——一个能够通过自然语言提示独立编写代码的智能体开发平台。瓦斯奎兹指出，该平台的推出旨在应对软件开发工具可扩展性方面日益扩大的鸿沟。
"这些工具虽然能生成代码，但开发者无法引导整个流程或确保其符合团队标准，"她说道，"我们希望将人工智能驱动软件开发中所有令人兴奋的元素与开发者真正需要的结构化框架相结合。"
亚马逊云科技还展示了其DevOps智能体与安全智能体，这些工具可在软件构建过程中用于诊断错误并扫描漏洞。
据亚马逊云科技称，这些更新的核心意图在于提升速度与效率。
"过去需要数年才能完成的工作，现在只需几天，甚至几分钟就能实现，"瓦斯奎兹表示。
英国二手车交易平台Motorway展示了Kiro的实际应用效果。由于团队对AI编码工具的需求日益增长，该公司引入了Kiro作为统一的集中式系统，既能加速流程，又不会削弱监管能力。
为此，Kiro在编写任何代码之前会先生成用户故事、验收标准、技术设计文档和架构图，构建一个引导代码开发的框架，而整个过程手动执行需要数天时间。
Motorway首席工程师瑞安·科马克告诉《AI商业》杂志，目前超过80%的Motorway工程师每天都在使用Kiro，该平台每月生成的代码量已超过100万行。
"我们并非要用AI彻底改变软件组织的工作方式，"科马克说，"而是用它来更快完成我们想做的事情。"
这种速度也伴随着风险——Kiro编写代码的速度已超过人类工程师可靠审查的速度。
"标准化流程对我们来说变得至关重要，因为不同团队对Kiro生成代码的审查标准各不相同，"科马克补充道，"我们确保在规划阶段建立了非常严谨的工程流程，同时让工程师引导Kiro进行代码编写，从而避免失去监管。"
随着人们对AI治理的担忧日益加剧，这种区分显得尤为重要。对此，科马克指出，Motorway采用了亚马逊云科技的共担责任模型来维护安全性，而Kiro强制性的规划阶段和审查节点则有助于确保大规模应用中的透明度。
展望未来，科马克认为该平台的潜力远未被充分挖掘。
"它问世还不到一年，但我们已经见证了行业内的爆炸性变革，"他说，"我们非常期待它还能带来哪些改变。"
亚马逊云科技EMEA地区市场拓展可持续发展负责人希拉里·谭告诉《AI商业》杂志，该公司也在探索环境监测领域的数据采集工作。
为此，亚马逊云科技与伦敦自然历史博物馆合作，在博物馆位于南肯辛顿的花园中部署了传感器网络，实时采集环境数据。
研究人员可以利用这些数据分析气温上升、重交通等城市状况如何影响生物多样性，并模拟潜在的缓解方案。谭将花园描述为亚马逊云科技的"首个活体实验室"，用于探索AI如何将环境数据转化为洞察。
"目前我们已经收集了约800万个数据点，而且这个数字每小时都在增长，"她说，"我们的最终目标是将其转化为对政策制定者和企业具有实际价值的洞察，帮助他们采取正确的干预措施，实现人与地球的更好平衡。"
谭表示，除直接应用外，该项目还体现了组织应如何重新思考可持续发展数据的更广泛转变。她认为，虽然欧洲ESG报告要求浪潮正推动运营数据向云基础设施迁移，但合规仅是起点。
"一旦拥有了数据基础，它还能告诉你什么？如何更好地服务客户？哪些地方存在创新机会？"她说，"我们正在将可持续发展从成本中心转变为孕育新商业模式的地方，在某些情况下，甚至是全新业务的诞生地。"

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.
The tech giant is positioning autonomous, long-running agents as the next defining shift in enterprise AI.
LONDON -- AWS has latched on firmly to the concept of frontier agents, systems that take AI agents a step beyond merely assistive tools to those that can complete complex tasks entirely autonomously.
In a keynote at the vendor’s London Summit this week, Francessca Vasquez, vice president of professional services and agentic AI at AWS, framed these systems around three core capabilities: autonomy, scale, and persistence.
“Frontier agents are a new class of agents that are significantly more capable,” Vasquez said. “You can direct them toward a goal, and they will figure out exactly how to achieve it. They're massively scaled, able to perform multiple concurrent tasks and capable of working for hours or even days in pursuit of ambitious and sometimes amorphous goals.”
In this landscape, AWS last year launched Kiro, an agentic development platform that independently writes code using natural language prompts. Vasquez framed the launch as addressing a widening gap in the scalability of software development tools.
“These tools were generating code, but builders couldn't guide the process or ensure it aligned with their team standards,” she said. “We wanted to take everything that is exciting about AI-powered software development and add the structure that our developers really need.”
AWS also showcased its DevOps Agent and Security Agent, which are used to diagnose errors and scan for vulnerabilities as software is being built.
The intention behind the updates is essentially about speed and efficiency, according to AWS
“What used to take years can now be done in days, if not minutes,” Vasquez said.
A concrete demonstration of Kiro in action came from U.K. used car marketplace Motorway. With its teams increasingly demanding AI coding tools, the company introduced Krio as a single, centralized system that could accelerate processes without jeopardizing oversight.
To this end, Kiro generates user stories, acceptance criteria, technical design documents, and architecture diagrams before any code is written, creating a framework to guide code development in a process that would take days to execute manually.
Ryan Cormack, principal engineer at Motorway, told AI Business that more than 80% of Motorway’s engineers are now daily users, with Kiro generating more than a million lines of code each month.
“We're not using AI to wildly change the way that we as a software organization are working,” Cormack said. “We're using it to just do things that we want to do quicker.”
That speed comes with its own risk; however, Kiro is capable of writing code faster than human engineers can reliably check.
“Again, standardizing processes became very important for us because our teams were all reviewing the Kiro-generated code differently,” Cormack added. “We made sure we have really strong engineering processes around the planning phase, and that engineers are steering Kiro through the code-writing, so we don’t lose oversight.”
That distinction is particularly relevant as concerns about AI governance grow. On this point, Cormack noted that Motorway used AWS’s shared responsibility model to maintain security, while Kiro’s inclusion of mandatory planning phases and review checkpoints helps ensure transparency at scale.
Looking ahead, Cormack says the platform's potential is still largely untapped.
“It’s not even a year old and we’ve already seen explosive change in the industry,” he said. “We're just really excited to see what else it can change.”
Hillary Tam, head of go-to-market sustainability for EMEA at AWS, told AI Business about how the vendor is also exploring data collection in environmental monitoring.
AWS has partnered with London’s Natural History Museum as part of this effort, deploying a network of sensors across the museum’s gardens in South Kensington to capture environmental data in real time.
Using the data, researchers can analyze how urban conditions such as rising temperatures and heavy traffic can affect biodiversity, and model potential mitigating solutions. Tam described the gardens as AWS’s “first living lab” exploring how AI can transform environmental data into insights.
“We’ve got around eight million data points now, and it’s growing by the hour,” she said. “Ultimately, we want to turn that into actionable insight for policymakers and businesses so they can make the right interventions that bring people and planet into better balance.”
Beyond the immediate application, Tam said the project illustrates a wider shift in how organizations should think about sustainability data. While a wave of European ESG reporting requirements is pushing operational data into cloud infrastructure, compliance, she argued, is just the starting point.
"Once you have that data foundation, what more can it tell you? How can you better serve your customers? Where are there opportunities for innovation?" She said, "We're moving sustainability from a cost center to a space where new business models, and in some cases entirely new businesses, are being born.”