<?xml version="1.0" encoding="UTF-8"?>
<rss version="2.0" xmlns:ve="https://vectorepoch.com/ns/intelligence">
  <channel>
    <title>向量纪元 AI 情报</title>
    <link>https://vectorepoch.com/intelligence/</link>
    <description>批准范围：日报与深度文章；历史资料未经重新核验</description>
    <language>zh-CN</language>
    <ve:releaseId>public-20261103T012715-bf933eaa</ve:releaseId>
    <item>
      <title>制造业工程Agent如何辅助替代件评审：证据组装与工程批准必须分开</title>
      <link>https://vectorepoch.com/intelligence/articles/engineering-agent-evidence-not-substitution-approval-20261103/</link>
      <guid>https://vectorepoch.com/intelligence/articles/engineering-agent-evidence-not-substitution-approval-20261103/</guid>
      <description>工程Agent更适合先承担资料检索、规格对照和评审草稿，而不是直接批准替代件。AWS的新教程提供了一个可拆解的设计示例，但其零件数据来自虚构制造商，不能当作真实航空交付结果。</description>
      <ve:id>ve-09e0f1e162fc4503b1fe5692f6be049e</ve:id>
      <ve:source url="https://aws.amazon.com/blogs/machine-learning/agentic-engineering-intelligence-with-amazon-nova-multimodal-embeddings-and-amazon-bedrock-agentcore">AWS Machine Learning Blog</ve:source>
      <ve:source url="https://developer.nvidia.com/blog/build-world-models-for-interactive-simulation-with-nvidia-cosmos">NVIDIA Technical Blog Feed</ve:source>
    </item>
    <item>
      <title>企业故障Agent如何验收：分诊提速不能当作自动修复成功</title>
      <link>https://vectorepoch.com/intelligence/articles/agent-triage-root-cause-and-human-remediation-20261103/</link>
      <guid>https://vectorepoch.com/intelligence/articles/agent-triage-root-cause-and-human-remediation-20261103/</guid>
      <description>Circle K公开案例和NVIDIA Attribution说明了两种证据组织方式：模型辅助汇集上下文，或确定性规则关联故障。企业应将诊断、批准与修复分别验收，避免把一条解释完整的消息当作事故已经解决。</description>
      <ve:id>ve-0d090210bfc54802b5bfc6f38a2bd927</ve:id>
      <ve:source url="https://aws.amazon.com/blogs/machine-learning/how-circle-k-accelerated-devops-incident-triage-to-under-15-minutes-with-amazon-bedrock">AWS Machine Learning Blog</ve:source>
      <ve:source url="https://developer.nvidia.com/blog/diagnose-distributed-ai-workload-failures-with-nvidia-attribution">NVIDIA Technical Blog Feed</ve:source>
    </item>
    <item>
      <title>企业知识库如何减少“抓到了却答不好”：正文、来源与任务评估分开验收</title>
      <link>https://vectorepoch.com/intelligence/articles/web-knowledge-preparation-and-agent-task-evaluation-20261103/</link>
      <guid>https://vectorepoch.com/intelligence/articles/web-knowledge-preparation-and-agent-task-evaluation-20261103/</guid>
      <description>网页入库只是知识库建设的一步。AWS的Crawl4AI教程强调正文准备和元数据，蒸馏教程强调成功任务轨迹，OpenAI的LifeVal则从实际任务出发。三者提示企业：需要验证可追溯的任务结果，而不是只统计页面数量。</description>
      <ve:id>ve-2489f1cfd3d445fdacc8710e2a3ad36e</ve:id>
      <ve:source url="https://aws.amazon.com/blogs/machine-learning/prepare-web-content-for-amazon-bedrock-knowledge-bases-with-crawl4ai">AWS Machine Learning Blog</ve:source>
      <ve:source url="https://aws.amazon.com/blogs/machine-learning/build-efficient-agents-with-knowledge-distillation-on-amazon-sagemaker-ai">AWS Machine Learning Blog</ve:source>
      <ve:source url="https://openai.com/index/lifeval">OpenAI News RSS</ve:source>
    </item>
    <item>
      <title>AI产业与企业智能化日报｜2026-11-03</title>
      <link>https://vectorepoch.com/intelligence/daily/2026-11-03/</link>
      <guid>https://vectorepoch.com/intelligence/daily/2026-11-03/</guid>
      <description>本期10条快讯、3篇分析，聚焦工程证据、故障分诊和知识库任务验收。来源首次发表日均为11月2日；Google条目为十月回顾，不将回顾日期当作产品新发布日。无足够独立GEO更新，本期不凑栏目。</description>
      <ve:id>ve-ed7a8853b6d941e491ab325814ade9b4</ve:id>
      <ve:source url="https://openai.com/index/lifeval">OpenAI News RSS</ve:source>
      <ve:source url="https://blog.google/innovation-and-ai/technology/ai/google-ai-updates-october-2026">Google AI / The Keyword RSS</ve:source>
      <ve:source url="https://aws.amazon.com/blogs/machine-learning/build-efficient-agents-with-knowledge-distillation-on-amazon-sagemaker-ai">AWS Machine Learning Blog</ve:source>
      <ve:source url="https://aws.amazon.com/blogs/machine-learning/agentic-engineering-intelligence-with-amazon-nova-multimodal-embeddings-and-amazon-bedrock-agentcore">AWS Machine Learning Blog</ve:source>
      <ve:source url="https://aws.amazon.com/blogs/machine-learning/how-circle-k-accelerated-devops-incident-triage-to-under-15-minutes-with-amazon-bedrock">AWS Machine Learning Blog</ve:source>
      <ve:source url="https://aws.amazon.com/blogs/machine-learning/prepare-web-content-for-amazon-bedrock-knowledge-bases-with-crawl4ai">AWS Machine Learning Blog</ve:source>
      <ve:source url="https://www.databricks.com/blog/lakebase-postgres-now-generally-available-google-cloud">Databricks Blog Feed</ve:source>
      <ve:source url="https://developer.nvidia.com/blog/build-world-models-for-interactive-simulation-with-nvidia-cosmos">NVIDIA Technical Blog Feed</ve:source>
      <ve:source url="https://developer.nvidia.com/blog/diagnose-distributed-ai-workload-failures-with-nvidia-attribution">NVIDIA Technical Blog Feed</ve:source>
      <ve:source url="https://openai.com/index/scaling-git-at-openai">OpenAI News RSS</ve:source>
    </item>
    <item>
      <title>语音Agent为何不能只看响应速度：识别、决策与音画同步要分别验收</title>
      <link>https://vectorepoch.com/intelligence/articles/voice-agent-modules-state-and-real-world-tests-20261101/</link>
      <guid>https://vectorepoch.com/intelligence/articles/voice-agent-modules-state-and-real-world-tests-20261101/</guid>
      <description>语音Agent是由识别、决策、发声和交互状态共同组成的系统。Hugging Face的新组件介绍与NVIDIA作者的MonsterHouse原型说明，统一接口和流畅演示不能替代真实噪声、打断、状态连续性及人工接管测试，尤其不能把合成场景当作现场结果。</description>
      <ve:id>ve-1b0a6b1193a246a5ba9976fb75e02636</ve:id>
      <ve:source url="https://huggingface.co/blog/anydecisionmodel">Hugging Face Blog Feed</ve:source>
      <ve:source url="https://huggingface.co/blog/nvidia/nemotron-halloween">Hugging Face Blog Feed</ve:source>
    </item>
    <item>
      <title>AI产业与企业智能化日报｜2026-11-01</title>
      <link>https://vectorepoch.com/intelligence/daily/2026-11-01/</link>
      <guid>https://vectorepoch.com/intelligence/daily/2026-11-01/</guid>
      <description>本期收录10条快讯与3篇分析，聚焦隔离试错、语音Agent分项验收和公开资料可复用性。原文日期为10月29日至31日；Google训练偏好公告为持续复核，不作新发布。制造业AI栏目本期无新增入选，厂商案例与编辑判断分开呈现。</description>
      <ve:id>ve-676e1bcecae548478e7dda1daff8fd53</ve:id>
      <ve:source url="https://www.databricks.com/blog/lakebase-branching-cicd-lessons-100-pr-databases">Databricks Blog Feed</ve:source>
      <ve:source url="https://www.databricks.com/blog/deliberate-decisions-and-enterprise-ai-value">Databricks Blog Feed</ve:source>
      <ve:source url="https://aws.amazon.com/blogs/machine-learning/configure-the-chatgpt-desktop-app-with-amazon-bedrock-in-aws-govcloud-us">AWS Machine Learning Blog</ve:source>
      <ve:source url="https://blogs.nvidia.com/blog/ai-joins-the-search-for-extraterrestrial-signals">NVIDIA Blog Feed</ve:source>
      <ve:source url="https://aws.amazon.com/blogs/machine-learning/accelerate-agentic-ai-with-nvidia-nemotron-models-on-amazon-eks-and-sagemaker-hyperpod-eks">AWS Machine Learning Blog</ve:source>
      <ve:source url="https://huggingface.co/blog/nvidia/nemotron-halloween">Hugging Face Blog Feed</ve:source>
      <ve:source url="https://huggingface.co/blog/mistral-large4">Hugging Face Blog Feed</ve:source>
      <ve:source url="https://huggingface.co/blog/anydecisionmodel">Hugging Face Blog Feed</ve:source>
      <ve:source url="https://huggingface.co/blog/announcing-project-protostar">Hugging Face Blog Feed</ve:source>
      <ve:source url="https://developers.google.com/search/blog/2026/10/ai-training-control">Google Search Central Blog Feed</ve:source>
    </item>
    <item>
      <title>公开资料怎样变得可供AI使用：ProtoStar与网站训练偏好的不同边界</title>
      <link>https://vectorepoch.com/intelligence/articles/ai-ready-content-training-preferences-and-evidence-20261101/</link>
      <guid>https://vectorepoch.com/intelligence/articles/ai-ready-content-training-preferences-and-evidence-20261101/</guid>
      <description>公开可访问、可供机器理解和允许特定用途，是三类不同条件。ProtoStar关注科研资源的文档、元数据与复用验证；Google的训练偏好计划关注网站表达使用偏好。两者都不能被简化为保证AI收录或推荐的开关。</description>
      <ve:id>ve-ddc4cab33533483b9c92f586715d40be</ve:id>
      <ve:source url="https://huggingface.co/blog/announcing-project-protostar">Hugging Face Blog Feed</ve:source>
      <ve:source url="https://developers.google.com/search/blog/2026/10/ai-training-control">Google Search Central Blog Feed</ve:source>
    </item>
    <item>
      <title>企业Agent如何安全试错：隔离测试数据不等于生产授权</title>
      <link>https://vectorepoch.com/intelligence/articles/agent-isolated-data-tests-and-production-readiness-20261101/</link>
      <guid>https://vectorepoch.com/intelligence/articles/agent-isolated-data-tests-and-production-readiness-20261101/</guid>
      <description>企业Agent的安全试错需要把可写测试环境、数据访问权限和生产变更批准分开。Databricks的Lakebase案例提供按任务隔离数据库的做法，其与DataIQ合作的报告则提醒：试点成功仍不能证明运行支持、业务责任和长期成本已经落实。</description>
      <ve:id>ve-ed9bb38a77d6405ca79f3cfa5cace275</ve:id>
      <ve:source url="https://www.databricks.com/blog/lakebase-branching-cicd-lessons-100-pr-databases">Databricks Blog Feed</ve:source>
      <ve:source url="https://www.databricks.com/blog/deliberate-decisions-and-enterprise-ai-value">Databricks Blog Feed</ve:source>
      <ve:source url="https://aws.amazon.com/blogs/machine-learning/configure-the-chatgpt-desktop-app-with-amazon-bedrock-in-aws-govcloud-us">AWS Machine Learning Blog</ve:source>
    </item>
    <item>
      <title>AI产业与企业智能化日报｜2026-10-31</title>
      <link>https://vectorepoch.com/intelligence/daily/2026-10-31/</link>
      <guid>https://vectorepoch.com/intelligence/daily/2026-10-31/</guid>
      <description>本期围绕Agent交付中的状态、权限与验收边界，收录10条快讯与3篇分析。来源原文日期均为10月30日；覆盖Agent、FDE、GEO、制造业AI和基础设施与治理。大模型厂商栏目本期留空，不用不可核验内容补齐。厂商计划、公开案例和编辑判断分别标识。</description>
      <ve:id>ve-133de2ccf18c4446a98039a2f5b3988a</ve:id>
      <ve:source url="https://developers.google.com/search/blog/2026/10/ai-training-control">Google Search Central Blog Feed</ve:source>
      <ve:source url="https://developer.nvidia.com/blog/build-specialized-agents-on-a-stack-you-control-with-nvidia-nemoclaw">NVIDIA Technical Blog Feed</ve:source>
      <ve:source url="https://aws.amazon.com/blogs/machine-learning/preview-web-apps-live-as-your-coding-agent-builds-them-with-amazon-bedrock-agentcore-runtime">AWS Machine Learning Blog</ve:source>
      <ve:source url="https://aws.amazon.com/blogs/machine-learning/how-worldwide-express-automated-freight-exceptions-with-amazon-bedrock">AWS Machine Learning Blog</ve:source>
      <ve:source url="https://www.databricks.com/blog/wealth-management-client-onboarding-needs-more-ai">Databricks Blog Feed</ve:source>
      <ve:source url="https://www.databricks.com/blog/databricks-and-sap-unleash-end-end-data-ai-capabilities-level-enterprise-ai-transformation">Databricks Blog Feed</ve:source>
      <ve:source url="https://developer.nvidia.com/blog/run-ai-models-for-edge-application-development-with-just-a-few-commands">NVIDIA Technical Blog Feed</ve:source>
      <ve:source url="https://aws.amazon.com/blogs/machine-learning/managing-amazon-bedrock-costs-end-to-end-a-playbook-for-finops-and-platform-teams">AWS Machine Learning Blog</ve:source>
      <ve:source url="https://developer.nvidia.com/blog/build-a-validated-kubernetes-runtime-for-nvidia-dynamo">NVIDIA Technical Blog Feed</ve:source>
      <ve:source url="https://developer.nvidia.com/blog/how-to-check-and-configure-nvidia-gpudirect-storage-for-your-platform">NVIDIA Technical Blog Feed</ve:source>
    </item>
    <item>
      <title>AI网页预览可用之后，还差哪些正式交付验收</title>
      <link>https://vectorepoch.com/intelligence/articles/agent-live-preview-production-acceptance-20261031/</link>
      <guid>https://vectorepoch.com/intelligence/articles/agent-live-preview-production-acceptance-20261031/</guid>
      <description>AI编程工具生成可操作预览有助于尽早发现问题，但会话预览、运行时检查和正式生产交付是不同阶段。AWS的AgentCore预览教程与NVIDIA的运行时、存储诊断指南，分别展示了访问期限、依赖检查和改动权限的边界。</description>
      <ve:id>ve-21e364440410435294a676c2aa2dfc73</ve:id>
      <ve:source url="https://aws.amazon.com/blogs/machine-learning/preview-web-apps-live-as-your-coding-agent-builds-them-with-amazon-bedrock-agentcore-runtime">AWS Machine Learning Blog</ve:source>
      <ve:source url="https://developer.nvidia.com/blog/build-a-validated-kubernetes-runtime-for-nvidia-dynamo">NVIDIA Technical Blog Feed</ve:source>
      <ve:source url="https://developer.nvidia.com/blog/how-to-check-and-configure-nvidia-gpudirect-storage-for-your-platform">NVIDIA Technical Blog Feed</ve:source>
    </item>
    <item>
      <title>企业Agent成本如何管理：先核对成功结果，再讨论调用价格</title>
      <link>https://vectorepoch.com/intelligence/articles/agent-cost-governance-outcome-boundaries-20261031/</link>
      <guid>https://vectorepoch.com/intelligence/articles/agent-cost-governance-outcome-boundaries-20261031/</guid>
      <description>企业Agent成本治理需要同时回答花费归属、执行边界和业务成功三个问题。AWS新指南把成本控制拆为五阶段；NVIDIA的运行时蓝图则说明，模型编排和访问控制不能混为一层。本文区分厂商方法与向量纪元实施建议，不提供收益承诺。</description>
      <ve:id>ve-df73b3162cae449a94eef48b4b7eef25</ve:id>
      <ve:source url="https://aws.amazon.com/blogs/machine-learning/managing-amazon-bedrock-costs-end-to-end-a-playbook-for-finops-and-platform-teams">AWS Machine Learning Blog</ve:source>
      <ve:source url="https://developer.nvidia.com/blog/build-specialized-agents-on-a-stack-you-control-with-nvidia-nemoclaw">NVIDIA Technical Blog Feed</ve:source>
    </item>
    <item>
      <title>企业Agent为什么需要共享业务状态：从物流异常到带版本的证据</title>
      <link>https://vectorepoch.com/intelligence/articles/enterprise-agent-shared-state-human-review-20261031/</link>
      <guid>https://vectorepoch.com/intelligence/articles/enterprise-agent-shared-state-human-review-20261031/</guid>
      <description>企业Agent交付的难点不只是把任务拆给多个模型，而是让后续动作基于当前有效证据，仍在原业务系统中完成审核。Worldwide Express的公开部署案例与Databricks的案件上下文方案提供了两种不同层级的材料：前者是案例方披露的实践，后者是厂商架构建议。</description>
      <ve:id>ve-e8b784b5e1034411ad5b481965d23b16</ve:id>
      <ve:source url="https://aws.amazon.com/blogs/machine-learning/how-worldwide-express-automated-freight-exceptions-with-amazon-bedrock">AWS Machine Learning Blog</ve:source>
      <ve:source url="https://www.databricks.com/blog/wealth-management-client-onboarding-needs-more-ai">Databricks Blog Feed</ve:source>
    </item>
    <item>
      <title>企业AI如何控制资源消耗：费用归因、访问限制与训练抢占不能混为一谈</title>
      <link>https://vectorepoch.com/intelligence/articles/serverless-cost-and-training-quota-controls-20261030/</link>
      <guid>https://vectorepoch.com/intelligence/articles/serverless-cost-and-training-quota-controls-20261030/</guid>
      <description>Serverless省去部分资源配置工作，并不自动消除超支风险。Databricks的费用控制更新与AWS Batch训练配额说明，企业需要分别管理谁能运行、费用归谁以及资源何时可借用。本文按正式可用、Beta和后续计划区分功能，不把预算提醒写成支出保证。</description>
      <ve:id>ve-16fd0fc4e7264216af1817346097bfbc</ve:id>
      <ve:source url="https://www.databricks.com/blog/smarter-cost-controls-serverless">Databricks 官方博客</ve:source>
      <ve:source url="https://aws.amazon.com/blogs/machine-learning/lend-and-borrow-resources-across-teams-with-aws-batchs-quota-management-for-sagemaker-training-jobs">AWS Machine Learning Blog</ve:source>
    </item>
    <item>
      <title>AI交付如何证明完成：从UI测试到代码迁移的验收器</title>
      <link>https://vectorepoch.com/intelligence/articles/ai-delivery-test-oracles-and-sample-gaps-20261030/</link>
      <guid>https://vectorepoch.com/intelligence/articles/ai-delivery-test-oracles-and-sample-gaps-20261030/</guid>
      <description>AI能执行步骤，不代表它能独立证明结果正确。Epsilon的UI测试实践与AWS的Kiro迁移教程分别展示了轨迹证据和行为一致性检查；后者也明确暴露样例尚未完成验收。对企业交付团队而言，最重要的是先定义证据，再讨论自动化比例。</description>
      <ve:id>ve-381b4273e87142948fc1f90926d69582</ve:id>
      <ve:source url="https://aws.amazon.com/blogs/machine-learning/building-intent-driven-ui-testing-with-amazon-nova-act">AWS Machine Learning Blog</ve:source>
      <ve:source url="https://aws.amazon.com/blogs/machine-learning/build-a-repeatable-php-to-java-migration-engine-with-kiro">AWS Machine Learning Blog</ve:source>
      <ve:source url="https://openai.com/index/ropes-gray-ai-diligence">OpenAI 官方案例</ve:source>
    </item>
    <item>
      <title>企业API如何变成可信Agent工具：MoEngage案例中的四条边界</title>
      <link>https://vectorepoch.com/intelligence/articles/enterprise-api-tools-platform-boundaries-20261030/</link>
      <guid>https://vectorepoch.com/intelligence/articles/enterprise-api-tools-platform-boundaries-20261030/</guid>
      <description>企业把API交给Agent，不是把完整接口目录直接放进对话。MoEngage公开案例将业务API、工具描述、执行身份和服务端权限分别管理，为已有大量系统接口的企业提供了可检查的工程参考。本文区分案例事实与向量纪元实施建议，不将其收益估计外推为通用结论。</description>
      <ve:id>ve-b5b0db998f204bc6ada4ad0caa2c0ace</ve:id>
      <ve:source url="https://aws.amazon.com/blogs/machine-learning/how-moengage-turned-over-400-apis-into-agent-tools-with-agentcore-gateway">AWS Machine Learning Blog</ve:source>
    </item>
    <item>
      <title>AI产业与企业智能化日报｜2026-10-30</title>
      <link>https://vectorepoch.com/intelligence/daily/2026-10-30/</link>
      <guid>https://vectorepoch.com/intelligence/daily/2026-10-30/</guid>
      <description>本期关注企业API工具化、独立验收与资源消耗边界。10条快讯和3篇分析依据10月29日一手原文；覆盖模型更新、Agent、FDE、制造业AI与基础设施治理，GEO没有合格新增事实，不凑稿。厂商效果、案例结果与编辑判断分别标明。</description>
      <ve:id>ve-ba468637a9cf4bc7888424bf999aa9dd</ve:id>
      <ve:source url="https://huggingface.co/blog/ibm-granite/granite-switch-4-1-1">IBM Granite 官方团队</ve:source>
      <ve:source url="https://openai.com/index/updated-model-spec">OpenAI 官方公告</ve:source>
      <ve:source url="https://aws.amazon.com/blogs/machine-learning/how-moengage-turned-over-400-apis-into-agent-tools-with-agentcore-gateway">AWS Machine Learning Blog</ve:source>
      <ve:source url="https://aws.amazon.com/blogs/machine-learning/building-intent-driven-ui-testing-with-amazon-nova-act">AWS Machine Learning Blog</ve:source>
      <ve:source url="https://aws.amazon.com/blogs/machine-learning/build-a-repeatable-php-to-java-migration-engine-with-kiro">AWS Machine Learning Blog</ve:source>
      <ve:source url="https://openai.com/index/ropes-gray-ai-diligence">OpenAI 官方案例</ve:source>
      <ve:source url="https://www.databricks.com/blog/announcing-general-availability-real-time-mode-spark-declarative-pipelines-databricks">Databricks 官方博客</ve:source>
      <ve:source url="https://www.databricks.com/blog/smarter-cost-controls-serverless">Databricks 官方博客</ve:source>
      <ve:source url="https://aws.amazon.com/blogs/machine-learning/lend-and-borrow-resources-across-teams-with-aws-batchs-quota-management-for-sagemaker-training-jobs">AWS Machine Learning Blog</ve:source>
      <ve:source url="https://developer.nvidia.com/blog/unlocking-tensor-core-performance-for-machine-learning-interatomic-potentials">NVIDIA Technical Blog</ve:source>
    </item>
    <item>
      <title>企业Agent如何从会话走到可信执行：权限、运行评估与退出机制</title>
      <link>https://vectorepoch.com/intelligence/articles/agent-business-permissions-runtime-evidence-20261029/</link>
      <guid>https://vectorepoch.com/intelligence/articles/agent-business-permissions-runtime-evidence-20261029/</guid>
      <description>向量纪元编辑判断：企业Agent的交付对象应是有权限边界、可核对结果、能安全退出的业务动作，而不只是一个会聊天的入口。ServiceNow合作公告与AWS平台参考蓝图提供了两种不同层级的证据，不能混成同一产品或同一上线承诺。</description>
      <ve:id>ve-3fe1dc0b3459488cbeec167cac0ea075</ve:id>
      <ve:source url="https://openai.com/index/openai-servicenow-partnership-expansion">OpenAI 官方公告</ve:source>
      <ve:source url="https://aws.amazon.com/blogs/machine-learning/building-an-agentic-ai-platform-on-amazon-bedrock-agentcore">AWS Machine Learning Blog</ve:source>
      <ve:source url="https://aws.amazon.com/blogs/machine-learning/accelerating-software-delivery-with-agentic-qa-automation-using-amazon-nova-act-part-3">AWS Machine Learning Blog</ve:source>
    </item>
    <item>
      <title>企业AI如何学会业务规则：从合成人物到稳定的数据契约</title>
      <link>https://vectorepoch.com/intelligence/articles/enterprise-ai-local-rules-data-contracts-20261029/</link>
      <guid>https://vectorepoch.com/intelligence/articles/enterprise-ai-local-rules-data-contracts-20261029/</guid>
      <description>向量纪元编辑判断：让模型熟悉某种语言，不等于让它掌握本地流程；让数据能够查询，也不等于让不同系统对同一字段有相同解释。NVIDIA本地化教程与Procore公开案例分别说明了回答依据和数据语义需要如何被维护。</description>
      <ve:id>ve-47b369f1ea704bb28af50d9886c119dc</ve:id>
      <ve:source url="https://developer.nvidia.com/blog/fine-tuning-nvidia-nemotron-to-know-local-rules-not-just-local-languages">NVIDIA Technical Blog</ve:source>
      <ve:source url="https://www.databricks.com/blog/how-procore-powering-construction-analytics-opensharing-0">Databricks 官方博客</ve:source>
    </item>
    <item>
      <title>语音Agent为什么不能只验音色：业务上下文、身份核对与人工接管</title>
      <link>https://vectorepoch.com/intelligence/articles/voice-agent-context-consent-handoff-20261029/</link>
      <guid>https://vectorepoch.com/intelligence/articles/voice-agent-context-consent-handoff-20261029/</guid>
      <description>向量纪元编辑判断：语音Agent能否投入企业服务，取决于是否知道当前通话对应谁、允许查询什么、何时必须转人工。Databricks的DIVA蓝图展示了这套分工，但它仍是厂商方案，不能据此承诺客服成本或满意度改善。</description>
      <ve:id>ve-bd67d82e46c648d2ae3ae0236a812eeb</ve:id>
      <ve:source url="https://www.databricks.com/blog/voice-agents-are-here-context-turns-them-real-customer-value">Databricks 官方博客</ve:source>
    </item>
    <item>
      <title>AI产业与企业智能化日报｜2026-10-29</title>
      <link>https://vectorepoch.com/intelligence/daily/2026-10-29/</link>
      <guid>https://vectorepoch.com/intelligence/daily/2026-10-29/</guid>
      <description>本期关注企业Agent的权限继承、语音上下文及业务规则维护。区分模型公告与实际开放、厂商参考实现与现场效果；FDE和GEO缺少合格新增事实，不为栏目齐全凑稿。制造业一条为已核实旧事件的持续观察。</description>
      <ve:id>ve-cb39178205bc4e308a98cdeeb6676a07</ve:id>
      <ve:source url="https://deepmind.google/blog/gemini-4-carbon-our-next-frontier-model">Google 官方博客</ve:source>
      <ve:source url="https://openai.com/index/openai-servicenow-partnership-expansion">OpenAI 官方公告</ve:source>
      <ve:source url="https://aws.amazon.com/blogs/machine-learning/building-an-agentic-ai-platform-on-amazon-bedrock-agentcore">AWS Machine Learning Blog</ve:source>
      <ve:source url="https://www.databricks.com/blog/voice-agents-are-here-context-turns-them-real-customer-value">Databricks 官方博客</ve:source>
      <ve:source url="https://developer.nvidia.com/blog/fine-tuning-nvidia-nemotron-to-know-local-rules-not-just-local-languages">NVIDIA Technical Blog</ve:source>
      <ve:source url="https://aws.amazon.com/blogs/machine-learning/distributed-training-with-ray-train-dlcs-and-efa-on-eks">AWS Machine Learning Blog</ve:source>
      <ve:source url="https://aws.amazon.com/blogs/machine-learning/accelerating-software-delivery-with-agentic-qa-automation-using-amazon-nova-act-part-3">AWS Machine Learning Blog</ve:source>
      <ve:source url="https://www.databricks.com/blog/how-procore-powering-construction-analytics-opensharing-0">Databricks 官方博客</ve:source>
      <ve:source url="https://openai.com/index/legora">OpenAI 官方案例</ve:source>
      <ve:source url="https://aws.amazon.com/blogs/machine-learning/fleet-wide-computer-vision-lifecycle-at-the-edge-with-sagemaker-sdk-v3">AWS Machine Learning Blog</ve:source>
    </item>
    <item>
      <title>工业端侧AI如何选择小模型：先限定任务，再验证证据与拒答</title>
      <link>https://vectorepoch.com/intelligence/articles/edge-ai-evidence-abstention-20261028/</link>
      <guid>https://vectorepoch.com/intelligence/articles/edge-ai-evidence-abstention-20261028/</guid>
      <description>工业端侧AI应先根据离线、延迟与任务范围确定约束，再比较模型大小。向量纪元依据Exowatt能源设备案例及Liquid AI的LFM3说明，讨论小模型专门化的适用条件。重点是数据、引用核验和人工接管，而不是用参数规模证明安全；厂商测试与研究发布不能替代企业现场、硬件和业务责任验收。</description>
      <ve:id>ve-2c9e1d2a92734c91a4943b6d9a49f033</ve:id>
      <ve:source url="https://aws.amazon.com/blogs/machine-learning/exowatt-beats-frontier-models-on-accuracy-with-sagemaker-serverless-model-customization">AWS 官方案例</ve:source>
      <ve:source url="https://huggingface.co/blog/LiquidAI/lfm3">Liquid AI 官方团队文章</ve:source>
    </item>
    <item>
      <title>文档抽取如何避免越改越错：字段、覆盖率与发布门槛一起验收</title>
      <link>https://vectorepoch.com/intelligence/articles/document-extraction-regression-coverage-20261028/</link>
      <guid>https://vectorepoch.com/intelligence/articles/document-extraction-regression-coverage-20261028/</guid>
      <description>文档抽取的改进应以整套字段和固定人工审核样本为单位，而不是修好一个错误就上线。向量纪元依据Databricks订单文档项目，面向制造企业资料处理与交付团队梳理规则、版本和测试的分工。厂商的改进比例只适用于其公开项目；企业仍需用自己的文档布局、缺失值规则与人工标注建立基线。</description>
      <ve:id>ve-55b01726249948e5997951bf80af16fd</ve:id>
      <ve:source url="https://www.databricks.com/blog/spot-fixes-measured-gains-improving-document-extraction-accuracy-aiextract-and-mlflow-prompt">Databricks 官方博客</ve:source>
    </item>
    <item>
      <title>企业AI评估先查标准答案：从Property Finder案例看交付证据</title>
      <link>https://vectorepoch.com/intelligence/articles/evaluation-label-bias-field-delivery-20261028/</link>
      <guid>https://vectorepoch.com/intelligence/articles/evaluation-label-bias-field-delivery-20261028/</guid>
      <description>企业比较AI方案前，应先检查标准答案是否独立于被测模型。向量纪元依据AWS与Property Finder共同公开的案例，说明标签偏差、测试集重叠与调优投入如何影响结论。本文面向企业评估与交付团队，关注可重复的方法，而不把房地产图片任务的精度或历史成本估算迁移成制造业收益承诺。</description>
      <ve:id>ve-7b96bb29d4ee4b25b498b69fb1dee5ae</ve:id>
      <ve:source url="https://aws.amazon.com/blogs/machine-learning/fix-evaluation-bias-to-unlock-better-models-on-amazon-bedrock">AWS 官方案例</ve:source>
      <ve:source url="https://www.databricks.com/blog/vital-skills-upskilling-genie-code-healthcare-and-life-sciences-skills">Databricks 官方博客</ve:source>
    </item>
    <item>
      <title>AI产业与企业智能化日报｜2026-10-28</title>
      <link>https://vectorepoch.com/intelligence/daily/2026-10-28/</link>
      <guid>https://vectorepoch.com/intelligence/daily/2026-10-28/</guid>
      <description>本期聚焦任务专门化、现场经验产品化与评估证据。纠正自动分类：世界模型研究预览不是制造业落地，联合查询教程不是FDE客户成果；GEO无足够已核实更新，不凑栏目。</description>
      <ve:id>ve-c08454a901a84e37bf36f4335098820e</ve:id>
      <ve:source url="https://huggingface.co/blog/LiquidAI/lfm3">Liquid AI 官方团队文章</ve:source>
      <ve:source url="https://deepmind.google/blog/genie-4-a-new-generation-of-world-simulation">Google DeepMind 官方博客</ve:source>
      <ve:source url="https://openai.com/index/oracle-apps-in-chatgpt">OpenAI 官方公告</ve:source>
      <ve:source url="https://www.databricks.com/blog/vital-skills-upskilling-genie-code-healthcare-and-life-sciences-skills">Databricks 官方博客</ve:source>
      <ve:source url="https://www.databricks.com/blog/spot-fixes-measured-gains-improving-document-extraction-accuracy-aiextract-and-mlflow-prompt">Databricks 官方博客</ve:source>
      <ve:source url="https://aws.amazon.com/blogs/machine-learning/fix-evaluation-bias-to-unlock-better-models-on-amazon-bedrock">AWS 官方案例</ve:source>
      <ve:source url="https://aws.amazon.com/blogs/machine-learning/exowatt-beats-frontier-models-on-accuracy-with-sagemaker-serverless-model-customization">AWS 官方案例</ve:source>
      <ve:source url="https://developer.nvidia.com/blog/memory-enforced-gpu-sharing-in-kubernetes-with-nvidia-cuda-13-4">NVIDIA Technical Blog</ve:source>
      <ve:source url="https://aws.amazon.com/blogs/machine-learning/announcing-soci-indexed-images-for-faster-sagemaker-ai-jobs">AWS 官方博客</ve:source>
      <ve:source url="https://www.databricks.com/blog/unlocking-investment-insights-without-migrating-your-data-lakehouse-federation">Databricks 官方博客</ve:source>
    </item>
    <item>
      <title>AI产业与企业智能化日报｜2026-10-27</title>
      <link>https://vectorepoch.com/intelligence/daily/2026-10-27/</link>
      <guid>https://vectorepoch.com/intelligence/daily/2026-10-27/</guid>
      <description>本期聚焦逐相机质检、企业知识库、Agent评估与业务成效边界。按正文纠正自动分类：企业AI Search不是GEO，连接器与教育教程不冒充FDE；NVIDIA原文首发时间与Feed修改时间分开记录。</description>
      <ve:id>ve-00085a5895f84efc8c9d4338d802752d</ve:id>
      <ve:source url="https://aws.amazon.com/blogs/machine-learning/fleet-wide-computer-vision-lifecycle-at-the-edge-with-sagemaker-sdk-v3">AWS Machine Learning Blog</ve:source>
      <ve:source url="https://aws.amazon.com/blogs/machine-learning/groundhog-day-for-ai-agents-prompt-caching-patterns-for-using-openai-models-on-amazon-bedrock">AWS Machine Learning Blog</ve:source>
      <ve:source url="https://aws.amazon.com/blogs/machine-learning/build-agentic-retrieval-on-a-managed-amazon-bedrock-knowledge-base-with-cdk">AWS Machine Learning Blog</ve:source>
      <ve:source url="https://www.databricks.com/blog/building-live-intelligence-databricks-ai-search-supports-streaming-tables-materialized-views">Databricks Blog Feed</ve:source>
      <ve:source url="https://www.databricks.com/blog/unify-your-productivity-and-ai-data-lakeflow-connect">Databricks Blog Feed</ve:source>
      <ve:source url="https://openai.com/index/chatgpt-expanding-world-of-work">OpenAI News RSS</ve:source>
      <ve:source url="https://openai.com/index/fanatics-streamline-fan-experience">OpenAI News RSS</ve:source>
      <ve:source url="https://openai.com/index/gpt-6-1-sol-coderabbit-code-security">OpenAI News RSS</ve:source>
      <ve:source url="https://developer.nvidia.com/blog/how-nvidia-mission-control-2-5-brings-autonomous-recovery-to-kubernetes">NVIDIA Technical Blog Feed</ve:source>
      <ve:source url="https://blog.google/products-and-platforms/products/education/ai-quests-polar-peaks">Google AI / The Keyword RSS</ve:source>
    </item>
    <item>
      <title>制造业视觉质检如何从单相机扩展到多产线：验收实际交付的模型产物</title>
      <link>https://vectorepoch.com/intelligence/articles/industrial-vision-per-camera-artifact-validation-20261027/</link>
      <guid>https://vectorepoch.com/intelligence/articles/industrial-vision-per-camera-artifact-validation-20261027/</guid>
      <description>制造业视觉质检扩展的关键不只是多训练几个模型，而是把每台相机的样本、质量标准、版本和设备产物对应起来。向量纪元依据AWS逐相机参考架构，说明为何模型评价、编译与登记需要分别留证。本文面向工厂技术与交付团队，讨论实施检查，不把合成数据演示当作现场质量保证。</description>
      <ve:id>ve-47efd2dab9f547589a1be7239b2e01cf</ve:id>
      <ve:source url="https://aws.amazon.com/blogs/machine-learning/fleet-wide-computer-vision-lifecycle-at-the-edge-with-sagemaker-sdk-v3">AWS Machine Learning Blog</ve:source>
    </item>
    <item>
      <title>企业如何读懂AI成效案例：任务样本、业务指标与人工批准不能省略</title>
      <link>https://vectorepoch.com/intelligence/articles/ai-business-case-task-metrics-human-review-20261027/</link>
      <guid>https://vectorepoch.com/intelligence/articles/ai-business-case-task-metrics-human-review-20261027/</guid>
      <description>企业判断AI应用价值，应先看任务范围和验证方法，再看增长数字。向量纪元结合Fanatics商品推荐、CodeRabbit安全审查与OpenAI工作活动数据，解释三个常见误读：互动提升不是成交提升，发现问题不是完成修复，个人使用分布不是企业部署规模。本文提供评估流程，不替代企业实际测试。</description>
      <ve:id>ve-8c7aea89976f44818519bf6f73a95962</ve:id>
      <ve:source url="https://openai.com/index/fanatics-streamline-fan-experience">OpenAI News RSS</ve:source>
      <ve:source url="https://openai.com/index/gpt-6-1-sol-coderabbit-code-security">OpenAI News RSS</ve:source>
      <ve:source url="https://openai.com/index/chatgpt-expanding-world-of-work">OpenAI News RSS</ve:source>
    </item>
    <item>
      <title>企业知识库怎样保持可用：接入、解析、检索和依据核验分开验收</title>
      <link>https://vectorepoch.com/intelligence/articles/enterprise-knowledge-ingestion-retrieval-evaluation-20261027/</link>
      <guid>https://vectorepoch.com/intelligence/articles/enterprise-knowledge-ingestion-retrieval-evaluation-20261027/</guid>
      <description>企业知识库可用性取决于资料是否进入、文本是否正确、索引是否更新及回答是否有依据，而非只看问答界面。向量纪元结合Databricks连接器和AI Search说明、AWS多知识库参考方案，面向企业实施团队梳理四段验收路径。托管服务减少部分维护工作，但不替企业决定哪些资料可以使用。</description>
      <ve:id>ve-eccab14f04fe48d09f25864317ee98d0</ve:id>
      <ve:source url="https://www.databricks.com/blog/unify-your-productivity-and-ai-data-lakeflow-connect">Databricks Blog Feed</ve:source>
      <ve:source url="https://www.databricks.com/blog/building-live-intelligence-databricks-ai-search-supports-streaming-tables-materialized-views">Databricks Blog Feed</ve:source>
      <ve:source url="https://aws.amazon.com/blogs/machine-learning/build-agentic-retrieval-on-a-managed-amazon-bedrock-knowledge-base-with-cdk">AWS Machine Learning Blog</ve:source>
    </item>
    <item>
      <title>企业Agent如何证明任务做对了：结构检查、结果测试与奖励设计</title>
      <link>https://vectorepoch.com/intelligence/articles/agent-validation-structure-outcome-reward-20261026/</link>
      <guid>https://vectorepoch.com/intelligence/articles/agent-validation-structure-outcome-reward-20261026/</guid>
      <description>企业Agent验收至少要区分能否执行、结果是否正确、评价目标是否合理。向量纪元结合Databricks声明式管线测试与AWS多轮强化学习教程，面向企业自动化与数据团队说明三项检查的职责：预运行校验能拦住部分结构错误，结果测试发现业务错误，奖励设计决定训练追求什么。厂商实验不等于企业效果保证。</description>
      <ve:id>ve-1645ba30c83a4093adf0c46e111dc09f</ve:id>
      <ve:source url="https://www.databricks.com/blog/built-agents-case-spark-declarative-pipelines">Databricks Blog Feed</ve:source>
      <ve:source url="https://aws.amazon.com/blogs/machine-learning/multi-turn-reinforcement-learning-on-amazon-sagemaker-ai-a-hands-on-guide">AWS Machine Learning Blog</ve:source>
    </item>
    <item>
      <title>Bayer案例说明什么：企业Agent入口不应取代受治理的执行系统</title>
      <link>https://vectorepoch.com/intelligence/articles/bayer-agent-governed-execution-and-delivery-20261026/</link>
      <guid>https://vectorepoch.com/intelligence/articles/bayer-agent-governed-execution-and-delivery-20261026/</guid>
      <description>Bayer的ADE是在既有CDE分析平台之上增加的自然语言入口，而不是替代执行平台。向量纪元依据AWS与Bayer公开案例，面向企业AI实施团队分析状态、身份、工具和任务监控的分工。其经验重点是把问题回答接到可核对的数据与运行记录上，公开案例中的收益数字不能直接作为其他企业的交付承诺。</description>
      <ve:id>ve-5aea4159c8d14c71a90635cd362cce9f</ve:id>
      <ve:source url="https://aws.amazon.com/blogs/machine-learning/how-bayer-crop-science-built-an-agentic-decision-engine-on-amazon-bedrock-agentcore-with-aws-unified-operations">AWS Machine Learning Blog</ve:source>
    </item>
    <item>
      <title>AI推理迁移如何验收：模型能运行、精度达标与性能改善不能混为一谈</title>
      <link>https://vectorepoch.com/intelligence/articles/ai-inference-compatibility-accuracy-performance-20261026/</link>
      <guid>https://vectorepoch.com/intelligence/articles/ai-inference-compatibility-accuracy-performance-20261026/</guid>
      <description>AI推理迁移需要依次确认兼容性、业务质量和实际负载表现，不能把“PASS”当作生产验收。向量纪元结合Hugging Face的Trainium支持查询、NVIDIA的TensorRT调优说明与AWS端点更新教程，面向企业技术团队提出分层检查方法：先限定测试范围，再核对精度，最后验证性能和运行依赖是否完整恢复。</description>
      <ve:id>ve-c9553214613743b6a9bbdf604323f401</ve:id>
      <ve:source url="https://huggingface.co/blog/torch-neuronx-model-support">Hugging Face Blog Feed</ve:source>
      <ve:source url="https://developer.nvidia.com/blog/scaling-nvidia-tensorrt-performance-optimization-beyond-a-single-build-route-with-the-global-performance-tuner-part-2">NVIDIA Technical Blog Feed</ve:source>
      <ve:source url="https://aws.amazon.com/blogs/machine-learning/orchestrating-sagemaker-ai-endpoint-auto-scaling-with-aws-cdk">AWS Machine Learning Blog</ve:source>
    </item>
    <item>
      <title>AI产业与企业智能化日报｜2026-10-26</title>
      <link>https://vectorepoch.com/intelligence/daily/2026-10-26/</link>
      <guid>https://vectorepoch.com/intelligence/daily/2026-10-26/</guid>
      <description>本期核验10月23日发布的官方技术材料与公开案例，聚焦Agent任务验收、企业执行边界、推理调优与机器人仿真。按正文纠正自动栏目：普通工程教程不冒充FDE，地图GeoJSON不属于GEO；本期大模型厂商、FDE与GEO栏目缺席。</description>
      <ve:id>ve-cd09459c29c34903a65dbbf35fccd92a</ve:id>
      <ve:source url="https://www.databricks.com/blog/built-agents-case-spark-declarative-pipelines">Databricks Blog Feed</ve:source>
      <ve:source url="https://aws.amazon.com/blogs/machine-learning/multi-turn-reinforcement-learning-on-amazon-sagemaker-ai-a-hands-on-guide">AWS Machine Learning Blog</ve:source>
      <ve:source url="https://aws.amazon.com/blogs/machine-learning/how-bayer-crop-science-built-an-agentic-decision-engine-on-amazon-bedrock-agentcore-with-aws-unified-operations">AWS Machine Learning Blog</ve:source>
      <ve:source url="https://huggingface.co/blog/allenai/litpivot">Ai2 / Hugging Face</ve:source>
      <ve:source url="https://developer.nvidia.com/blog/using-ai-agents-to-teach-robots-like-microduck-to-walk-in-simulation">NVIDIA Technical Blog Feed</ve:source>
      <ve:source url="https://developer.nvidia.com/blog/scaling-nvidia-tensorrt-performance-optimization-beyond-a-single-build-route-with-the-global-performance-tuner-part-2">NVIDIA Technical Blog Feed</ve:source>
      <ve:source url="https://huggingface.co/blog/torch-neuronx-model-support">Hugging Face Blog Feed</ve:source>
      <ve:source url="https://aws.amazon.com/blogs/machine-learning/orchestrating-sagemaker-ai-endpoint-auto-scaling-with-aws-cdk">AWS Machine Learning Blog</ve:source>
      <ve:source url="https://www.databricks.com/blog/real-time-energy-trading-and-analytics-databricks">Databricks Blog Feed</ve:source>
      <ve:source url="https://blogs.nvidia.com/blog/kazakhstan-ai-compute-ecosystem">NVIDIA Blog Feed</ve:source>
    </item>
    <item>
      <title>企业Agent如何从回答走向执行：原始依据、隔离测试与人工批准</title>
      <link>https://vectorepoch.com/intelligence/articles/enterprise-agent-action-testing-human-approval-20261021/</link>
      <guid>https://vectorepoch.com/intelligence/articles/enterprise-agent-action-testing-human-approval-20261021/</guid>
      <description>企业Agent执行是否完成，应由可核对的业务记录和验收结果证明，而不是由一句“已处理”证明。向量纪元结合供应链参考方案、Smartsheet接入说明和Hamlet公开案例，梳理面向企业实施团队的执行边界：先获得依据，再隔离验证，最后由有权限的人批准高影响动作。</description>
      <ve:id>ve-4087315648f543de8cf1a14f1df98e91</ve:id>
      <ve:source url="https://aws.amazon.com/blogs/machine-learning/from-data-to-decisions-conversational-supply-chain-intelligence-with-amazon-quick-and-amazon-bedrock-agentcore">AWS Machine Learning Blog</ve:source>
      <ve:source url="https://aws.amazon.com/blogs/machine-learning/work-management-context-for-ai-agents-with-smartsheet-mcp-on-aws">AWS Machine Learning Blog</ve:source>
      <ve:source url="https://www.databricks.com/blog/how-hamlet-uses-databricks-lakebase-postgres-test-changes-against-production-data">Databricks Blog Feed</ve:source>
    </item>
    <item>
      <title>企业Agent如何记住信息：长期记忆、固定状态与可核验证据的区别</title>
      <link>https://vectorepoch.com/intelligence/articles/agent-memory-fixed-state-and-evidence-20261021/</link>
      <guid>https://vectorepoch.com/intelligence/articles/agent-memory-fixed-state-and-evidence-20261021/</guid>
      <description>企业Agent的记忆不是把历史对话无限保留。长期记忆负责提炼可复用信息，模型运行状态影响计算资源，原始证据负责支持业务结论，三者不能互相替代。向量纪元结合Databricks与i14/AWS公开材料，面向知识服务和长文档应用团队，说明应该分别验收什么；本刊未进行性能复测。</description>
      <ve:id>ve-a34a2ed58b2b430991146cd172656d77</ve:id>
      <ve:source url="https://www.databricks.com/blog/introducing-agent-memory-databricks">Databricks Blog Feed</ve:source>
      <ve:source url="https://aws.amazon.com/blogs/machine-learning/beyond-the-kv-cache-training-and-serving-i14s-laplace-3b-linear-attention-model-on-aws-trainium2-with-native-pytorch">AWS Machine Learning Blog</ve:source>
    </item>
    <item>
      <title>企业Agent接入工具前要确认什么：发现、身份与业务口径三道检查</title>
      <link>https://vectorepoch.com/intelligence/articles/agent-discovery-identity-and-semantic-governance-20261021/</link>
      <guid>https://vectorepoch.com/intelligence/articles/agent-discovery-identity-and-semantic-governance-20261021/</guid>
      <description>工具能够被搜索到，不意味着Agent已经有权执行，也不意味着不同系统的指标口径相同。向量纪元依据AWS工具发现、身份验证、Denodo语义层材料及Eon公开案例，将企业接入拆成能力发现、权限验证和业务语义检查，适用于需要连接多个系统的实施团队。</description>
      <ve:id>ve-a38f16f71d8c4fd4992d1ef51bcfd51f</ve:id>
      <ve:source url="https://aws.amazon.com/blogs/machine-learning/discover-tools-at-runtime-with-aws-agent-registry">AWS Machine Learning Blog</ve:source>
      <ve:source url="https://aws.amazon.com/blogs/machine-learning/securing-ai-agent-identity-with-x-509-pki-on-aws">AWS Machine Learning Blog</ve:source>
      <ve:source url="https://aws.amazon.com/blogs/machine-learning/build-a-semantic-layer-for-agentic-ai-on-aws-with-denodo-and-amazon-bedrock-agentcore">AWS Machine Learning Blog</ve:source>
      <ve:source url="https://aws.amazon.com/blogs/machine-learning/how-eon-built-multi-cloud-investigation-with-an-mcp-gateway-on-amazon-bedrock-agentcore">AWS Machine Learning Blog</ve:source>
    </item>
    <item>
      <title>AI产业与企业智能化日报｜2026-10-21</title>
      <link>https://vectorepoch.com/intelligence/daily/2026-10-21/</link>
      <guid>https://vectorepoch.com/intelligence/daily/2026-10-21/</guid>
      <description>本期聚焦Agent记忆、工具发现、身份与语义治理，以及供应链执行和工业AI合作。技术示例、客户自述和编辑判断分别标注；FDE与GEO无足够直接更新，不以一般开发工具或营销内容凑栏目。</description>
      <ve:id>ve-d531c8ee98ea430b9146b5a7ec7d370c</ve:id>
      <ve:source url="https://aws.amazon.com/blogs/machine-learning/beyond-the-kv-cache-training-and-serving-i14s-laplace-3b-linear-attention-model-on-aws-trainium2-with-native-pytorch">AWS Machine Learning Blog</ve:source>
      <ve:source url="https://aws.amazon.com/blogs/machine-learning/work-management-context-for-ai-agents-with-smartsheet-mcp-on-aws">AWS Machine Learning Blog</ve:source>
      <ve:source url="https://aws.amazon.com/blogs/machine-learning/from-data-to-decisions-conversational-supply-chain-intelligence-with-amazon-quick-and-amazon-bedrock-agentcore">AWS Machine Learning Blog</ve:source>
      <ve:source url="https://www.databricks.com/blog/introducing-agent-memory-databricks">Databricks Blog Feed</ve:source>
      <ve:source url="https://aws.amazon.com/blogs/machine-learning/discover-tools-at-runtime-with-aws-agent-registry">AWS Machine Learning Blog</ve:source>
      <ve:source url="https://aws.amazon.com/blogs/machine-learning/securing-ai-agent-identity-with-x-509-pki-on-aws">AWS Machine Learning Blog</ve:source>
      <ve:source url="https://aws.amazon.com/blogs/machine-learning/how-eon-built-multi-cloud-investigation-with-an-mcp-gateway-on-amazon-bedrock-agentcore">AWS Machine Learning Blog</ve:source>
      <ve:source url="https://aws.amazon.com/blogs/machine-learning/build-a-semantic-layer-for-agentic-ai-on-aws-with-denodo-and-amazon-bedrock-agentcore">AWS Machine Learning Blog</ve:source>
      <ve:source url="https://www.databricks.com/blog/how-hamlet-uses-databricks-lakebase-postgres-test-changes-against-production-data">Databricks Blog Feed</ve:source>
      <ve:source url="https://blogs.nvidia.com/blog/germany-wants-to-lead-the-world-in-industrial-ai-how-nvidia-will-help">NVIDIA Blog Feed</ve:source>
    </item>
    <item>
      <title>企业AI治理如何落到流程：受控访问、证据检查与原子发布</title>
      <link>https://vectorepoch.com/intelligence/articles/enterprise-ai-governance-atomic-publishing-20261019/</link>
      <guid>https://vectorepoch.com/intelligence/articles/enterprise-ai-governance-atomic-publishing-20261019/</guid>
      <description>企业AI治理需要将访问控制、质量检查和发布完整性转化为可验收步骤。向量纪元依据Itaú公开案例和Databricks事务材料说明这三类责任：治理平台约束使用范围，质量检查判断结果是否可信，事务保证相关写入共同完成；三者不能互相替代。</description>
      <ve:id>ve-07c960f349014ebf86c116686aca31e5</ve:id>
      <ve:source url="https://openai.com/index/itau-generative-ai">OpenAI News RSS</ve:source>
      <ve:source url="https://www.databricks.com/blog/transactions-are-generally-available-advancing-warehousing-lakehouse">Databricks Blog Feed</ve:source>
    </item>
    <item>
      <title>AI产业与企业智能化日报｜2026-10-19</title>
      <link>https://vectorepoch.com/intelligence/daily/2026-10-19/</link>
      <guid>https://vectorepoch.com/intelligence/daily/2026-10-19/</guid>
      <description>本期关注开放模型部署、本地Agent开发、知识产品接入与企业治理。Beam为跨期后续观察，未重复计算两个来源；制造业AI候选的Feed时间与原文首次日期不一致，未作为当日新消息。FDE和GEO无足够直接证据，不以平台更新凑栏目。厂商测试与客户自述不等于独立验证。</description>
      <ve:id>ve-18e81d30194f401e9e8ad84e3ca4891b</ve:id>
      <ve:source url="https://blogs.nvidia.com/blog/local-ai-rtx-spark-windows-laptops-celebration">NVIDIA Blog Feed</ve:source>
      <ve:source url="https://developer.nvidia.com/blog/code-test-and-deploy-on-one-machine-with-nvidia-rtx-spark">NVIDIA Technical Blog Feed</ve:source>
      <ve:source url="https://www.databricks.com/blog/transactions-are-generally-available-advancing-warehousing-lakehouse">Databricks Blog Feed</ve:source>
      <ve:source url="https://www.databricks.com/blog/closing-binary-blind-spot-domain-knowledge-injection-llm-malware-triage-npm-packages">Databricks Blog Feed</ve:source>
      <ve:source url="https://openai.com/index/itau-generative-ai">OpenAI News RSS</ve:source>
      <ve:source url="https://openai.com/index/stack-overflow-chatgpt-plugin">OpenAI News RSS</ve:source>
      <ve:source url="https://openai.com/index/benefit-street-partners">OpenAI News RSS</ve:source>
      <ve:source url="https://www.anthropic.com/news/public-benefit-fund">Anthropic News</ve:source>
      <ve:source url="https://huggingface.co/blog/nvidia/scaling-indic-pretraining">Hugging Face Blog Feed</ve:source>
      <ve:source url="https://aws.amazon.com/blogs/machine-learning/introducing-reflection-beam-an-open-weight-model-for-agentic-coding-on-aws/">AWS Machine Learning Blog</ve:source>
      <ve:source url="https://huggingface.co/blog/reflection-beam">Hugging Face Blog Feed</ve:source>
    </item>
    <item>
      <title>开放模型与本地AI怎么选：先区分部署控制、内存和业务适配</title>
      <link>https://vectorepoch.com/intelligence/articles/open-model-local-ai-deployment-boundaries-20261019/</link>
      <guid>https://vectorepoch.com/intelligence/articles/open-model-local-ai-deployment-boundaries-20261019/</guid>
      <description>开放权重和本地设备分别回答“谁控制运行环境”和“在哪里执行”，不能互相替代。向量纪元结合Beam与RTX Spark官方材料，整理面向企业技术团队的评估流程：先定任务，再核对资源、软件栈和真实负载；不把两者写成已经验证兼容的组合。</description>
      <ve:id>ve-1b8b6c11e7e5407886a475ebf75a3b51</ve:id>
      <ve:source url="https://aws.amazon.com/blogs/machine-learning/introducing-reflection-beam-an-open-weight-model-for-agentic-coding-on-aws/">AWS Machine Learning Blog</ve:source>
      <ve:source url="https://developer.nvidia.com/blog/code-test-and-deploy-on-one-machine-with-nvidia-rtx-spark">NVIDIA Technical Blog Feed</ve:source>
      <ve:source url="https://huggingface.co/blog/reflection-beam">Hugging Face Blog Feed</ve:source>
    </item>
    <item>
      <title>企业Agent如何验收：从知识接入、业务初稿到人工复核</title>
      <link>https://vectorepoch.com/intelligence/articles/enterprise-agent-evidence-review-and-adoption-20261019/</link>
      <guid>https://vectorepoch.com/intelligence/articles/enterprise-agent-evidence-review-and-adoption-20261019/</guid>
      <description>企业Agent验收应分开检查资料接入、结果可追溯、人工复核和使用入口。Stack Overflow与Benefit Street Partners的公开案例提供了流程样本，但案例自述的开发速度和工时不能作为其他企业的交付承诺。本文由向量纪元编辑，适用于需要知识检索、资料整理与辅助分析的企业团队。</description>
      <ve:id>ve-9705893a12994c2088eb8fd5e7efbccd</ve:id>
      <ve:source url="https://openai.com/index/stack-overflow-chatgpt-plugin">OpenAI News RSS</ve:source>
      <ve:source url="https://openai.com/index/benefit-street-partners">OpenAI News RSS</ve:source>
    </item>
    <item>
      <title>企业AI交付从模型选择延伸到部署、复用与领域工作流</title>
      <link>https://vectorepoch.com/intelligence/articles/enterprise-ai-open-model-delivery/</link>
      <guid>https://vectorepoch.com/intelligence/articles/enterprise-ai-open-model-delivery/</guid>
      <description>AWS、NVIDIA和Databricks的官方内容显示，企业AI工程正在同时关注开放模型部署、领域训练方法和可复用的交付资料。模型本身只是能力的一部分，数据、评测、运行环境和工作流程决定了最终能否稳定使用。</description>
      <ve:id>ve-41de5ff255834d67aa5f5757108a9bcf</ve:id>
      <ve:source url="https://aws.amazon.com/blogs/machine-learning/introducing-reflection-beam-an-open-weight-model-for-agentic-coding-on-aws/">AWS Machine Learning Blog</ve:source>
      <ve:source url="https://developer.nvidia.com/blog/efficient-moe-training-for-biological-foundation-models/">NVIDIA Technical Blog</ve:source>
      <ve:source url="https://www.databricks.com/blog/sizing-lakebase-search-lessons-real-app-and-4-million-passages">Databricks Blog</ve:source>
    </item>
    <item>
      <title>智能体进入生产后，评测与运行时控制为何必须同时建设</title>
      <link>https://vectorepoch.com/intelligence/articles/agent-evaluation-runtime-control/</link>
      <guid>https://vectorepoch.com/intelligence/articles/agent-evaluation-runtime-control/</guid>
      <description>AWS和NVIDIA近期官方技术文章分别从持续评测、分层控制和运行时监测出发，说明企业智能体要从可演示走向可交付，必须同时处理质量回退、权限边界和异常追踪。</description>
      <ve:id>ve-5889ccdb3e774320a7b33d5b02fd18c0</ve:id>
      <ve:source url="https://aws.amazon.com/blogs/machine-learning/building-trust-in-ai-agents-a-production-ready-evaluation-framework-with-amazon-bedrock-agentcore/">AWS Machine Learning Blog</ve:source>
      <ve:source url="https://aws.amazon.com/blogs/machine-learning/trust-in-depth-giving-agents-more-autonomy-without-losing-control/">AWS Machine Learning Blog</ve:source>
      <ve:source url="https://developer.nvidia.com/blog/nvidia-open-agent-safety-platform-a-reference-for-continuous-in-silicon-agent-monitoring/">NVIDIA Technical Blog</ve:source>
    </item>
    <item>
      <title>AI工厂基础设施从“有GPU”走向“工作负载可用”</title>
      <link>https://vectorepoch.com/intelligence/articles/ai-factory-workload-validation/</link>
      <guid>https://vectorepoch.com/intelligence/articles/ai-factory-workload-validation/</guid>
      <description>NVIDIA近期围绕GPU集群就绪度、Run:ai开源和ROS 2链路优化发布的官方内容共同指向一个工程事实：算力基础设施的价值必须通过真实任务、调度、网络、存储和应用链路验证。</description>
      <ve:id>ve-e50021f50adf4bb9b74bfad2784c7408</ve:id>
      <ve:source url="https://developer.nvidia.com/blog/validate-gpu-cluster-readiness-before-ai-workloads-land/">NVIDIA Technical Blog</ve:source>
      <ve:source url="https://developer.nvidia.com/blog/nvidia-runai-is-now-open-source-as-part-of-nvidia-dsx-ai-factory-platform/">NVIDIA Technical Blog</ve:source>
      <ve:source url="https://developer.nvidia.com/blog/accelerating-a-ros-2-node-with-an-ai-agent-and-nvidia-isaac-ros/">NVIDIA Technical Blog</ve:source>
    </item>
    <item>
      <title>AI产业与企业智能化日报｜2026-10-18</title>
      <link>https://vectorepoch.com/intelligence/daily/2026-10-18/</link>
      <guid>https://vectorepoch.com/intelligence/daily/2026-10-18/</guid>
      <description>本期基于已实际访问并核验的官方来源，关注智能体评测与运行时控制、AI工厂基础设施验收、开放模型部署，以及制造业机器人链路优化。今日未形成可核验的GEO栏目事件；FDE栏目采用平台工程与交付复用相关的一手技术内容，不将厂商案例收益外推为普遍结论。</description>
      <ve:id>ve-f5acfdc5b60e4781be6d758d423e706a</ve:id>
      <ve:source url="https://aws.amazon.com/blogs/machine-learning/building-trust-in-ai-agents-a-production-ready-evaluation-framework-with-amazon-bedrock-agentcore/">AWS Machine Learning Blog</ve:source>
      <ve:source url="https://aws.amazon.com/blogs/machine-learning/trust-in-depth-giving-agents-more-autonomy-without-losing-control/">AWS Machine Learning Blog</ve:source>
      <ve:source url="https://developer.nvidia.com/blog/add-runtime-controls-to-ai-agents-with-nvidia-openshell/">NVIDIA Technical Blog</ve:source>
      <ve:source url="https://developer.nvidia.com/blog/nvidia-nemotron-3-5-super-vl-powers-specialized-agents-with-multimodal-reasoning/">NVIDIA Technical Blog</ve:source>
      <ve:source url="https://aws.amazon.com/blogs/machine-learning/introducing-reflection-beam-an-open-weight-model-for-agentic-coding-on-aws/">AWS Machine Learning Blog</ve:source>
      <ve:source url="https://developer.nvidia.com/blog/efficient-moe-training-for-biological-foundation-models/">NVIDIA Technical Blog</ve:source>
      <ve:source url="https://developer.nvidia.com/blog/accelerating-a-ros-2-node-with-an-ai-agent-and-nvidia-isaac-ros/">NVIDIA Technical Blog</ve:source>
      <ve:source url="https://developer.nvidia.com/blog/nvidia-runai-is-now-open-source-as-part-of-nvidia-dsx-ai-factory-platform/">NVIDIA Technical Blog</ve:source>
      <ve:source url="https://developer.nvidia.com/blog/validate-gpu-cluster-readiness-before-ai-workloads-land/">NVIDIA Technical Blog</ve:source>
      <ve:source url="https://developer.nvidia.com/blog/nvidia-open-agent-safety-platform-a-reference-for-continuous-in-silicon-agent-monitoring/">NVIDIA Technical Blog</ve:source>
      <ve:source url="https://www.databricks.com/blog/how-our-ai-research-team-uses-ai-runtime">Databricks Blog</ve:source>
      <ve:source url="https://www.databricks.com/blog/sizing-lakebase-search-lessons-real-app-and-4-million-passages">Databricks Blog</ve:source>
    </item>
    <item>
      <title>AI工厂基础设施从“有GPU”走向“工作负载可用”</title>
      <link>https://vectorepoch.com/intelligence/articles/ai-factory-infrastructure-validation/</link>
      <guid>https://vectorepoch.com/intelligence/articles/ai-factory-infrastructure-validation/</guid>
      <description>NVIDIA近期围绕GPU集群就绪度、Run:ai开源和机器人ROS 2链路优化发布了多篇官方技术内容。它们共同指向一个工程事实：算力基础设施的价值要通过真实任务、调度、网络、存储和应用链路验证，而不是用单项硬件健康指标替代。</description>
      <ve:id>ve-0a49f61780cb4acabbcb70362e7e83da</ve:id>
      <ve:source url="https://developer.nvidia.com/blog/validate-gpu-cluster-readiness-before-ai-workloads-land/">NVIDIA Technical Blog</ve:source>
      <ve:source url="https://developer.nvidia.com/blog/nvidia-runai-is-now-open-source-as-part-of-nvidia-dsx-ai-factory-platform/">NVIDIA Technical Blog</ve:source>
      <ve:source url="https://developer.nvidia.com/blog/accelerating-a-ros-2-node-with-an-ai-agent-and-nvidia-isaac-ros/">NVIDIA Technical Blog</ve:source>
    </item>
    <item>
      <title>智能体进入生产后，评测与运行时治理为何必须同时建设</title>
      <link>https://vectorepoch.com/intelligence/articles/agent-evaluation-runtime-governance/</link>
      <guid>https://vectorepoch.com/intelligence/articles/agent-evaluation-runtime-governance/</guid>
      <description>近期AWS和NVIDIA的官方技术文章分别从持续评测与运行时安全两个方向，展示了企业智能体从开发到生产需要的质量控制方法。对企业而言，重点不是追逐“自主”叙事，而是把可验证的任务质量、权限边界和异常追踪写进交付标准。</description>
      <ve:id>ve-3d4b0256384f4a97ac187086af25d312</ve:id>
      <ve:source url="https://aws.amazon.com/blogs/machine-learning/building-trust-in-ai-agents-a-production-ready-evaluation-framework-with-amazon-bedrock-agentcore/">AWS Machine Learning Blog</ve:source>
      <ve:source url="https://developer.nvidia.com/blog/nvidia-open-agent-safety-platform-a-reference-for-continuous-in-silicon-agent-monitoring/">NVIDIA Technical Blog</ve:source>
      <ve:source url="https://developer.nvidia.com/blog/nvidia-nemotron-3-5-super-vl-powers-specialized-agents-with-multimodal-reasoning/">NVIDIA Technical Blog</ve:source>
    </item>
    <item>
      <title>企业AI竞争从模型发布延伸到训练、部署与领域工作流</title>
      <link>https://vectorepoch.com/intelligence/articles/enterprise-ai-model-and-platform-engineering/</link>
      <guid>https://vectorepoch.com/intelligence/articles/enterprise-ai-model-and-platform-engineering/</guid>
      <description>NVIDIA、Google DeepMind和Databricks的官方内容显示，企业AI工程正在同时关注高效训练、科学研究工作流和模型部署复现。模型本身只是能力的一部分，数据、评测、运行环境和业务流程决定了最终能否稳定使用。</description>
      <ve:id>ve-b07336b5284b4ee69f47b0d0bdee381c</ve:id>
      <ve:source url="https://developer.nvidia.com/blog/efficient-moe-training-for-biological-foundation-models/">NVIDIA Technical Blog</ve:source>
      <ve:source url="https://deepmind.google/blog/accelerating-antimicrobial-discovery/">Google DeepMind Blog RSS</ve:source>
      <ve:source url="https://www.databricks.com/blog/how-factset-cut-model-training-and-deployment-days-hours-databricks-ai-runtime">Databricks Blog</ve:source>
    </item>
    <item>
      <title>AI产业与企业智能化日报｜2026-10-17</title>
      <link>https://vectorepoch.com/intelligence/daily/2026-10-17/</link>
      <guid>https://vectorepoch.com/intelligence/daily/2026-10-17/</guid>
      <description>本期基于已核验的官方来源，关注智能体持续评测与运行时治理、AI工厂基础设施验收，以及模型训练和企业部署工程化。公开资料显示，企业AI的竞争焦点正在从单个模型能力扩展到可验证的生产流程；本期未形成足够高置信度的GEO栏目事件。</description>
      <ve:id>ve-f0b1298b819641e482ec005bd1a47c96</ve:id>
      <ve:source url="https://developer.nvidia.com/blog/efficient-moe-training-for-biological-foundation-models/">NVIDIA Technical Blog</ve:source>
      <ve:source url="https://deepmind.google/blog/accelerating-antimicrobial-discovery/">Google DeepMind Blog RSS</ve:source>
      <ve:source url="https://aws.amazon.com/blogs/machine-learning/building-trust-in-ai-agents-a-production-ready-evaluation-framework-with-amazon-bedrock-agentcore/">AWS Machine Learning Blog</ve:source>
      <ve:source url="https://developer.nvidia.com/blog/nvidia-nemotron-3-5-super-vl-powers-specialized-agents-with-multimodal-reasoning/">NVIDIA Technical Blog</ve:source>
      <ve:source url="https://aws.amazon.com/blogs/machine-learning/introducing-reflection-beam-an-open-weight-model-for-agentic-coding-on-aws/">AWS Machine Learning Blog</ve:source>
      <ve:source url="https://developer.nvidia.com/blog/nvidia-open-agent-safety-platform-a-reference-for-continuous-in-silicon-agent-monitoring/">NVIDIA Technical Blog</ve:source>
      <ve:source url="https://www.databricks.com/blog/how-our-ai-research-team-uses-ai-runtime">Databricks Blog</ve:source>
      <ve:source url="https://www.databricks.com/blog/how-factset-cut-model-training-and-deployment-days-hours-databricks-ai-runtime">Databricks Blog</ve:source>
      <ve:source url="https://developer.nvidia.com/blog/accelerating-a-ros-2-node-with-an-ai-agent-and-nvidia-isaac-ros/">NVIDIA Technical Blog</ve:source>
      <ve:source url="https://developer.nvidia.com/blog/nvidia-runai-is-now-open-source-as-part-of-nvidia-dsx-ai-factory-platform/">NVIDIA Technical Blog</ve:source>
      <ve:source url="https://developer.nvidia.com/blog/validate-gpu-cluster-readiness-before-ai-workloads-land/">NVIDIA Technical Blog</ve:source>
      <ve:source url="https://www.databricks.com/blog/reducing-noise-how-atlassian-built-ocsf-security-lakehouse-databricks">Databricks Blog</ve:source>
    </item>
    <item>
      <title>AI 工厂基础设施从“有算力”转向“可用、可测、可治理”</title>
      <link>https://vectorepoch.com/intelligence/articles/ai-factory-throughput-and-readiness/</link>
      <guid>https://vectorepoch.com/intelligence/articles/ai-factory-throughput-and-readiness/</guid>
      <description>NVIDIA与Databricks的官方材料把 AI 基础设施的关注点推进到资源调度、功耗效率、集群就绪和安全数据治理。对企业实施而言，算力采购只有在真实任务、数据链路和审计责任都能通过验收时才具有业务意义。</description>
      <ve:id>ve-292ef0a0cdcd44d5aaf7501d058407a8</ve:id>
      <ve:source url="https://developer.nvidia.com/blog/nvidia-runai-is-now-open-source-as-part-of-nvidia-dsx-ai-factory-platform/">NVIDIA Technical Blog</ve:source>
      <ve:source url="https://developer.nvidia.com/blog/how-nvidia-dsx-maxlps-maximizes-ai-factory-throughput-and-efficiency/">NVIDIA Technical Blog</ve:source>
      <ve:source url="https://developer.nvidia.com/blog/validate-gpu-cluster-readiness-before-ai-workloads-land/">NVIDIA Technical Blog</ve:source>
      <ve:source url="https://www.databricks.com/blog/reducing-noise-how-atlassian-built-ocsf-security-lakehouse-databricks">Databricks Blog</ve:source>
    </item>
    <item>
      <title>AI产业与企业智能化日报｜2026-10-16</title>
      <link>https://vectorepoch.com/intelligence/daily/2026-10-16/</link>
      <guid>https://vectorepoch.com/intelligence/daily/2026-10-16/</guid>
      <description>本期整理的官方材料覆盖 Agent 工作流、企业 AI 研究与检索、AI 工厂基础设施和 AI 搜索主题分析，来源首发时间并不都在当日。编辑判断：企业智能化应从模型接入推进到工作流验证、运行环境治理与可追溯数据。</description>
      <ve:id>ve-6927797d1240402e8717b30389d10b19</ve:id>
      <ve:source url="https://developer.nvidia.com/blog/efficient-moe-training-for-biological-foundation-models/">NVIDIA Technical Blog</ve:source>
      <ve:source url="https://deepmind.google/blog/accelerating-antimicrobial-discovery/">Google DeepMind Blog RSS</ve:source>
      <ve:source url="https://developer.nvidia.com/blog/nvidia-nemotron-3-5-super-vl-powers-specialized-agents-with-multimodal-reasoning/">NVIDIA Technical Blog</ve:source>
      <ve:source url="https://developer.nvidia.com/blog/scale-agentic-coding-workflows-with-metas-muse-spark-open-on-nvidia-gb300-nvl72/">NVIDIA Technical Blog</ve:source>
      <ve:source url="https://developer.nvidia.com/blog/manage-kubernetes-node-fleets-with-nodewright/">NVIDIA Technical Blog</ve:source>
      <ve:source url="https://www.databricks.com/blog/how-our-ai-research-team-uses-ai-runtime">Databricks Blog</ve:source>
      <ve:source url="https://www.databricks.com/blog/geospatial-queries-are-38x-faster-lakehousert">Databricks Blog</ve:source>
      <ve:source url="https://www.databricks.com/blog/sizing-lakebase-search-lessons-real-app-and-4-million-passages">Databricks Blog</ve:source>
      <ve:source url="https://blogs.bing.com/webmaster/2026/10/introducing-topics-portfolio-beta-in-ai-performance/">Bing Webmaster Blog</ve:source>
      <ve:source url="https://developer.nvidia.com/blog/nvidia-runai-is-now-open-source-as-part-of-nvidia-dsx-ai-factory-platform/">NVIDIA Technical Blog</ve:source>
      <ve:source url="https://developer.nvidia.com/blog/how-nvidia-dsx-maxlps-maximizes-ai-factory-throughput-and-efficiency/">NVIDIA Technical Blog</ve:source>
      <ve:source url="https://blogs.nvidia.com/blog/european-cities-agents-urban-operations/">NVIDIA Blog</ve:source>
      <ve:source url="https://www.databricks.com/blog/reducing-noise-how-atlassian-built-ocsf-security-lakehouse-databricks">Databricks Blog</ve:source>
      <ve:source url="https://developer.nvidia.com/blog/validate-gpu-cluster-readiness-before-ai-workloads-land/">NVIDIA Technical Blog</ve:source>
      <ve:source url="https://developer.nvidia.com/blog/introducing-nv-reason-ct-open-3d-ct-vlm-for-radiologist-chain-of-thought-reasoning/">NVIDIA Technical Blog</ve:source>
      <ve:source url="https://developer.nvidia.com/blog/nvidia-open-agent-safety-platform-a-reference-for-continuous-in-silicon-agent-monitoring/">NVIDIA Technical Blog</ve:source>
    </item>
    <item>
      <title>企业 AI 交付的两条底线：可验证的研究流程与可理解的知识结构</title>
      <link>https://vectorepoch.com/intelligence/articles/ai-research-and-searchable-enterprise-knowledge/</link>
      <guid>https://vectorepoch.com/intelligence/articles/ai-research-and-searchable-enterprise-knowledge/</guid>
      <description>Databricks 的 AI Research 与 Lakebase Search 内容，以及 Bing 的 Topics Portfolio 测试，共同指向实验评估、知识检索和 AI 搜索可见度的协同。</description>
      <ve:id>ve-915351d5310d4426820c4bd8add4bc0a</ve:id>
      <ve:source url="https://www.databricks.com/blog/how-our-ai-research-team-uses-ai-runtime">Databricks Blog</ve:source>
      <ve:source url="https://www.databricks.com/blog/sizing-lakebase-search-lessons-real-app-and-4-million-passages">Databricks Blog</ve:source>
      <ve:source url="https://blogs.bing.com/webmaster/2026/10/introducing-topics-portfolio-beta-in-ai-performance/">Bing Webmaster Blog</ve:source>
    </item>
    <item>
      <title>Agent竞争进入工作流阶段：从多模态推理到运行时控制</title>
      <link>https://vectorepoch.com/intelligence/articles/agent-workflow-from-model-to-production/</link>
      <guid>https://vectorepoch.com/intelligence/articles/agent-workflow-from-model-to-production/</guid>
      <description>本组官方材料显示，企业 Agent 的竞争重点正在从模型回答扩展到多模态输入、编码工作流、节点管理和运行时治理。对企业而言，真正重要的是这些能力能否在真实流程中被组合、验证并持续追踪。</description>
      <ve:id>ve-a2e815f39a004b63b74ccf584dc50dc3</ve:id>
      <ve:source url="https://developer.nvidia.com/blog/nvidia-nemotron-3-5-super-vl-powers-specialized-agents-with-multimodal-reasoning/">NVIDIA Technical Blog</ve:source>
      <ve:source url="https://developer.nvidia.com/blog/scale-agentic-coding-workflows-with-metas-muse-spark-open-on-nvidia-gb300-nvl72/">NVIDIA Technical Blog</ve:source>
      <ve:source url="https://developer.nvidia.com/blog/manage-kubernetes-node-fleets-with-nodewright/">NVIDIA Technical Blog</ve:source>
      <ve:source url="https://developer.nvidia.com/blog/nvidia-open-agent-safety-platform-a-reference-for-continuous-in-silicon-agent-monitoring/">NVIDIA Technical Blog</ve:source>
    </item>
    <item>
      <title>企业 Agent 进入运行治理阶段：文档、工具调用与持续监控要一起验收</title>
      <link>https://vectorepoch.com/intelligence/articles/enterprise-agent-runtime-and-document-governance-2026-10-15/</link>
      <guid>https://vectorepoch.com/intelligence/articles/enterprise-agent-runtime-and-document-governance-2026-10-15/</guid>
      <description>本期官方内容覆盖 Agent 安全监控、文档处理和云安全修复。企业 Agent 的验收标准正在从回答质量扩展到上下文、权限、异常处理、监控和人工接管。</description>
      <ve:id>ve-13a0ac1518f84cf8a68372f8ca3a8580</ve:id>
      <ve:source url="https://developer.nvidia.com/blog/nvidia-open-agent-safety-platform-a-reference-for-continuous-in-silicon-agent-monitoring/">NVIDIA Technical Blog</ve:source>
      <ve:source url="https://aws.amazon.com/blogs/machine-learning/build-self-improving-document-processing-using-amazon-bedrock-agentcore/">AWS Machine Learning Blog</ve:source>
      <ve:source url="https://aws.amazon.com/blogs/machine-learning/how-tamnoon-built-ai-powered-cloud-security-remediation-with-amazon-bedrock/">AWS Machine Learning Blog</ve:source>
    </item>
    <item>
      <title>向量纪元 AI 产业与企业智能化日报｜2026-10-15</title>
      <link>https://vectorepoch.com/intelligence/daily/2026-10-15/</link>
      <guid>https://vectorepoch.com/intelligence/daily/2026-10-15/</guid>
      <description>本期基于已访问且可访问的官方公开页面，覆盖大模型厂商、Agent、FDE、GEO、制造业AI和基础设施与治理六类动态。内容区分来源事实、厂商或案例方声明与向量纪元编辑判断，不把单一案例外推为普遍结论。</description>
      <ve:id>ve-54865b2dfc9141e09425420072f09f9f</ve:id>
      <ve:source url="https://aws.amazon.com/blogs/machine-learning/deploy-and-evaluate-models-with-amazon-sagemaker-jumpstart/">AWS Machine Learning Blog</ve:source>
      <ve:source url="https://www.databricks.com/blog/reducing-qwen-training-time-6x-ai-runtime">Databricks Blog</ve:source>
      <ve:source url="https://deepmind.google/blog/on-device-ai-scam-detection-mddi/">Google DeepMind Blog RSS</ve:source>
      <ve:source url="https://developer.nvidia.com/blog/nvidia-open-agent-safety-platform-a-reference-for-continuous-in-silicon-agent-monitoring/">NVIDIA Technical Blog</ve:source>
      <ve:source url="https://aws.amazon.com/blogs/machine-learning/build-self-improving-document-processing-using-amazon-bedrock-agentcore/">AWS Machine Learning Blog</ve:source>
      <ve:source url="https://aws.amazon.com/blogs/machine-learning/how-tamnoon-built-ai-powered-cloud-security-remediation-with-amazon-bedrock/">AWS Machine Learning Blog</ve:source>
      <ve:source url="https://www.databricks.com/blog/sizing-lakebase-search-lessons-real-app-and-4-million-passages">Databricks Blog</ve:source>
      <ve:source url="https://www.databricks.com/blog/introducing-private-network-gateway">Databricks Blog</ve:source>
      <ve:source url="https://blogs.bing.com/webmaster/2026/10/introducing-topics-portfolio-beta-in-ai-performance/">Bing Webmaster Blog</ve:source>
      <ve:source url="https://developer.nvidia.com/blog/nvidia-runai-is-now-open-source-as-part-of-nvidia-dsx-ai-factory-platform/">NVIDIA Technical Blog</ve:source>
      <ve:source url="https://blogs.nvidia.com/blog/european-cities-agents-urban-operations/">NVIDIA Blog</ve:source>
      <ve:source url="https://aws.amazon.com/blogs/machine-learning/accessing-openai-gpt-models-on-amazon-bedrock-with-global-cross-region-inference-in-thailand-malaysia-singapore-indonesia-and-taiwan/">AWS Machine Learning Blog</ve:source>
      <ve:source url="https://aws.amazon.com/blogs/machine-learning/run-multi-node-inference-workloads-using-ray-llm-deep-learning-containers/">AWS Machine Learning Blog</ve:source>
      <ve:source url="https://blogs.nvidia.com/blog/ocp-global-summit-vera-rubin-scale-ai-factories/">NVIDIA Blog</ve:source>
    </item>
    <item>
      <title>制造业 AI 先补工程基础：平台、数字孪生与现场边界要分层</title>
      <link>https://vectorepoch.com/intelligence/articles/manufacturing-ai-platform-and-digital-twin-2026-10-15/</link>
      <guid>https://vectorepoch.com/intelligence/articles/manufacturing-ai-platform-and-digital-twin-2026-10-15/</guid>
      <description>NVIDIA 的 Run:ai 开源、城市运营 Agent 与 AI 工厂平台内容显示，复杂场景需要计算基础设施、数据、仿真和系统集成共同工作。制造企业应先选择可留痕、可验收的工程环节。</description>
      <ve:id>ve-afa4e1ccb9044a9c8f583f0b4da228c7</ve:id>
      <ve:source url="https://developer.nvidia.com/blog/nvidia-runai-is-now-open-source-as-part-of-nvidia-dsx-ai-factory-platform/">NVIDIA Technical Blog</ve:source>
      <ve:source url="https://blogs.nvidia.com/blog/european-cities-agents-urban-operations/">NVIDIA Blog</ve:source>
      <ve:source url="https://blogs.nvidia.com/blog/ocp-global-summit-vera-rubin-scale-ai-factories/">NVIDIA Blog</ve:source>
    </item>
    <item>
      <title>GEO 进入主题化测量阶段：工具指标仍不能替代真实问答证据</title>
      <link>https://vectorepoch.com/intelligence/articles/geo-topic-measurement-and-evidence-2026-10-15/</link>
      <guid>https://vectorepoch.com/intelligence/articles/geo-topic-measurement-and-evidence-2026-10-15/</guid>
      <description>Bing Webmaster Tools 发布 Topics Portfolio Beta，用主题规模、被引用网站和机会查询观察 AI Search 生态。它为 GEO 提供了新的发现维度，但企业仍需回到真实问答、页面抓取和来源证据。</description>
      <ve:id>ve-ee037a2f1fd44c2bb91659e4326a7ede</ve:id>
      <ve:source url="https://blogs.bing.com/webmaster/2026/10/introducing-topics-portfolio-beta-in-ai-performance/">Bing Webmaster Blog</ve:source>
    </item>
    <item>
      <title>企业 Agent 正进入工具链阶段：从知识库、查询到成本和权限治理</title>
      <link>https://vectorepoch.com/intelligence/articles/enterprise-agent-toolchain-governance-2026-10-14/</link>
      <guid>https://vectorepoch.com/intelligence/articles/enterprise-agent-toolchain-governance-2026-10-14/</guid>
      <description>本期 AWS 官方内容同时覆盖企业 IT 服务台、自然语言安全查询、Bedrock 模型接入和多节点推理。企业 Agent 的判断标准正在从单次回答质量扩展到知识库新鲜度、工具权限、成本、延迟和失败回滚。</description>
      <ve:id>ve-52eab2bf5c3f4a6480ba8c8576108c29</ve:id>
      <ve:source url="https://aws.amazon.com/blogs/machine-learning/how-shipstation-global-built-an-ai-it-help-desk-on-amazon-bedrock-agentcore/">AWS Machine Learning Blog</ve:source>
      <ve:source url="https://aws.amazon.com/blogs/machine-learning/natural-language-security-queries-at-orca-security-using-amazon-bedrock/">AWS Machine Learning Blog</ve:source>
      <ve:source url="https://aws.amazon.com/blogs/machine-learning/run-multi-node-inference-workloads-using-ray-llm-deep-learning-containers/">AWS Machine Learning Blog</ve:source>
      <ve:source url="https://aws.amazon.com/blogs/machine-learning/accessing-openai-gpt-models-on-amazon-bedrock-with-global-cross-region-inference-in-thailand-malaysia-singapore-indonesia-and-taiwan/">AWS Machine Learning Blog</ve:source>
      <ve:source url="https://aws.amazon.com/blogs/machine-learning/manage-private-pricing-offers-for-amazon-bedrock-with-managed-entitlements/">AWS Machine Learning Blog</ve:source>
    </item>
    <item>
      <title>向量纪元 AI 产业与企业智能化日报｜2026-10-14</title>
      <link>https://vectorepoch.com/intelligence/daily/2026-10-14/</link>
      <guid>https://vectorepoch.com/intelligence/daily/2026-10-14/</guid>
      <description>本期基于已访问的官方公开页面，覆盖大模型厂商、Agent、FDE、GEO、制造业AI和基础设施与治理六类动态。内容严格区分来源事实、厂商声明与向量纪元判断；案例中的性能、效果和可用范围不外推为普遍结论。</description>
      <ve:id>ve-d872546224f546b48de42e81b283fcb3</ve:id>
      <ve:source url="https://blog.google/innovation-and-ai/models-and-research/gemini-models/introducing-gemini-4-argon/">Google Blog</ve:source>
      <ve:source url="https://blog.google/innovation-and-ai/technology/safety-security/gemini-4-model-safety-security/">Google Blog</ve:source>
      <ve:source url="https://aws.amazon.com/blogs/machine-learning/mixture-of-experts-kernel-engineering-on-amazon-sagemaker-ai/">AWS Machine Learning Blog</ve:source>
      <ve:source url="https://aws.amazon.com/blogs/machine-learning/manage-private-pricing-offers-for-amazon-bedrock-with-managed-entitlements/">AWS Machine Learning Blog</ve:source>
      <ve:source url="https://aws.amazon.com/blogs/machine-learning/how-shipstation-global-built-an-ai-it-help-desk-on-amazon-bedrock-agentcore/">AWS Machine Learning Blog</ve:source>
      <ve:source url="https://aws.amazon.com/blogs/machine-learning/natural-language-security-queries-at-orca-security-using-amazon-bedrock/">AWS Machine Learning Blog</ve:source>
      <ve:source url="https://huggingface.co/blog/LiquidAI/lfm25-24b-a2b">Liquid AI 团队（Hugging Face）</ve:source>
      <ve:source url="https://www.databricks.com/blog/practical-guide-real-time-analytics-2026">Databricks Blog</ve:source>
      <ve:source url="https://www.databricks.com/blog/whats-new-lakeflow-designer-october-2026-roundup">Databricks Blog</ve:source>
      <ve:source url="https://blogs.bing.com/webmaster/2026/10/introducing-topics-portfolio-beta-in-ai-performance/">Bing Webmaster Blog</ve:source>
      <ve:source url="https://developer.nvidia.com/blog/nvidia-runai-is-now-open-source-as-part-of-nvidia-dsx-ai-factory-platform/">NVIDIA Technical Blog</ve:source>
      <ve:source url="https://developers.googleblog.com/gemini-4-argon-for-robotics/">Google Developers Blog</ve:source>
      <ve:source url="https://blogs.nvidia.com/blog/local-ai-rtx-spark-laptops-software-updates/">NVIDIA Blog</ve:source>
      <ve:source url="https://aws.amazon.com/blogs/machine-learning/run-multi-node-inference-workloads-using-ray-llm-deep-learning-containers/">AWS Machine Learning Blog</ve:source>
      <ve:source url="https://aws.amazon.com/blogs/machine-learning/accessing-openai-gpt-models-on-amazon-bedrock-with-global-cross-region-inference-in-thailand-malaysia-singapore-indonesia-and-taiwan/">AWS Machine Learning Blog</ve:source>
      <ve:source url="https://developer.nvidia.com/blog/scaling-decision-optimization-to-100-million-variables-and-beyond-with-mpdlp-in-nvidia-cuopt/">NVIDIA Technical Blog</ve:source>
    </item>
    <item>
      <title>AI 可见度开始进入主题化测量：工具指标不能替代真实问答证据</title>
      <link>https://vectorepoch.com/intelligence/articles/ai-visibility-topic-measurement-boundary/</link>
      <guid>https://vectorepoch.com/intelligence/articles/ai-visibility-topic-measurement-boundary/</guid>
      <description>Bing Webmaster Tools 发布 Topics Portfolio Beta，用主题规模、被引用网站和机会查询观察 AI Search 生态。它为 GEO 工作提供了新的观察维度，但页面已明确提示分类和生态指标是近似值，企业仍需回到真实问答和公开证据。</description>
      <ve:id>ve-e8b42d6979b0449aa378c29a8f9a85d6</ve:id>
      <ve:source url="https://blogs.bing.com/webmaster/2026/10/introducing-topics-portfolio-beta-in-ai-performance/">Bing Webmaster Blog</ve:source>
    </item>
    <item>
      <title>制造业 AI 如何从模型发布走到工程流程：先做可验证的中间环节</title>
      <link>https://vectorepoch.com/intelligence/articles/manufacturing-ai-from-model-to-engineering-workflow/</link>
      <guid>https://vectorepoch.com/intelligence/articles/manufacturing-ai-from-model-to-engineering-workflow/</guid>
      <description>NVIDIA 和 Google DeepMind 的公开内容分别涉及 AI 工厂平台开源与机器人场景。对制造企业而言，落地重点不是把模型直接接到设备，而是先把资料、仿真、工程分析和安全边界做成可验收流程。</description>
      <ve:id>ve-eac6ea6e45994b53ba1739b82ef70be2</ve:id>
      <ve:source url="https://developer.nvidia.com/blog/nvidia-runai-is-now-open-source-as-part-of-nvidia-dsx-ai-factory-platform/">NVIDIA Technical Blog</ve:source>
      <ve:source url="https://developers.googleblog.com/gemini-4-argon-for-robotics/">Google Developers Blog</ve:source>
    </item>
    <item>
      <title>向量纪元 AI 产业与企业智能化日报｜2026-10-11</title>
      <link>https://vectorepoch.com/intelligence/daily/2026-10-11/</link>
      <guid>https://vectorepoch.com/intelligence/daily/2026-10-11/</guid>
      <description>本期聚焦 Agent 的评测与支付基础设施、GPU 资源治理、工业仿真和 AI 安全治理。正式内容仅采用已访问的官方公开页面；厂商案例、性能与投入金额均保留来源归因，向量纪元判断和企业建议单独呈现。今日可核验更新暂未覆盖 FDE 与 GEO 栏目。</description>
      <ve:id>ve-5ccb720bb6084737953c1eeed37aae04</ve:id>
      <ve:source url="https://aws.amazon.com/blogs/machine-learning/icymi-what-landed-for-ai-builders-in-september-2026/">AWS Machine Learning Blog</ve:source>
      <ve:source url="https://aws.amazon.com/blogs/machine-learning/pay-per-inference-for-ai-agents-how-blockrun-and-incarna-use-amazon-bedrock-agentcore-payments/">AWS Machine Learning Blog</ve:source>
      <ve:source url="https://developer.nvidia.com/blog/building-reliable-data-analytics-agents-lessons-from-the-kdd-cup/">NVIDIA Technical Blog</ve:source>
      <ve:source url="https://www.databricks.com/blog/biomedical-imagings-real-bottleneck-data-not-model">Databricks Blog</ve:source>
      <ve:source url="https://www.anthropic.com/news/genesis-mission-commitment">Anthropic News</ve:source>
      <ve:source url="https://www.anthropic.com/news/anthropic-cyber-mission">Anthropic News</ve:source>
      <ve:source url="https://developer.nvidia.com/blog/scale-bitwise-deterministic-pretraining-with-nvidia-megatron-core/">NVIDIA Technical Blog</ve:source>
      <ve:source url="https://aws.amazon.com/blogs/machine-learning/share-gpu-clusters-across-teams-with-isolation-and-fairness-using-amazon-sagemaker-hyperpod/">AWS Machine Learning Blog</ve:source>
      <ve:source url="https://blogs.nvidia.com/blog/geforce-now-thursday-gears-of-war-e-day/">NVIDIA Blog</ve:source>
      <ve:source url="https://blogs.nvidia.com/blog/developers-simulation-frontier-ai-agents/">NVIDIA Blog</ve:source>
    </item>
    <item>
      <title>工业 AI Agent 如何进入仿真流程？先验证资产与边界，再谈现场自动化</title>
      <link>https://vectorepoch.com/intelligence/articles/industrial-ai-agent-simulation-boundary/</link>
      <guid>https://vectorepoch.com/intelligence/articles/industrial-ai-agent-simulation-boundary/</guid>
      <description>NVIDIA 的公开内容展示了前沿 Agent 与 Omniverse 仿真工具结合的路径。对制造企业来说，真正可交付的重点不是让模型直接控制设备，而是把资产、物理假设、工具调用、测试轨迹和安全验收分层记录。</description>
      <ve:id>ve-8e481381d2bc49889f590d614f96ff79</ve:id>
      <ve:source url="https://blogs.nvidia.com/blog/developers-simulation-frontier-ai-agents/">NVIDIA Blog</ve:source>
    </item>
    <item>
      <title>GPU 基础设施治理的三个抓手：可复现训练、租户隔离与公平调度</title>
      <link>https://vectorepoch.com/intelligence/articles/gpu-governance-reproducible-training/</link>
      <guid>https://vectorepoch.com/intelligence/articles/gpu-governance-reproducible-training/</guid>
      <description>本期官方更新显示，GPU 平台治理正在从单纯扩容转向可复现训练、团队隔离、公平调度和成本归属。企业要把这些能力纳入平台验收，而不是等到账单或故障发生后再补治理。</description>
      <ve:id>ve-bedfd81692de4e04936c37bbb25396f3</ve:id>
      <ve:source url="https://developer.nvidia.com/blog/scale-bitwise-deterministic-pretraining-with-nvidia-megatron-core/">NVIDIA Technical Blog</ve:source>
      <ve:source url="https://aws.amazon.com/blogs/machine-learning/share-gpu-clusters-across-teams-with-isolation-and-fairness-using-amazon-sagemaker-hyperpod/">AWS Machine Learning Blog</ve:source>
    </item>
    <item>
      <title>Agent 平台进入系统工程阶段：从评测、知识库同步到支付限额</title>
      <link>https://vectorepoch.com/intelligence/articles/agent-platform-evaluation-payment-governance/</link>
      <guid>https://vectorepoch.com/intelligence/articles/agent-platform-evaluation-payment-governance/</guid>
      <description>AWS 近期官方内容同时覆盖 Agent 平台更新、知识库连接器、人工审批工作流和按请求支付案例。对企业而言，Agent 的可用性不应只看模型回答，而要看数据是否新鲜、工具是否受控、支出是否可审计以及失败时能否回退。</description>
      <ve:id>ve-fdc3ca71f36d47f1a642d074a6cbe959</ve:id>
      <ve:source url="https://aws.amazon.com/blogs/machine-learning/icymi-what-landed-for-ai-builders-in-september-2026/">AWS Machine Learning Blog</ve:source>
      <ve:source url="https://aws.amazon.com/blogs/machine-learning/pay-per-inference-for-ai-agents-how-blockrun-and-incarna-use-amazon-bedrock-agentcore-payments/">AWS Machine Learning Blog</ve:source>
    </item>
    <item>
      <title>Anthropic 2026 使用政策更新：企业 Agent 要补哪些人工与物理安全控制</title>
      <link>https://vectorepoch.com/intelligence/articles/anthropic-2026-usage-policy-enterprise-controls/</link>
      <guid>https://vectorepoch.com/intelligence/articles/anthropic-2026-usage-policy-enterprise-controls/</guid>
      <description>Anthropic 更新使用政策，重点澄清欺骗活动、武器、监控、执法、高风险建议和自主物理动作的边界。对企业 Agent 而言，最直接的实施要求是：影响健康、法律权利、财务、就业或基本服务的建议必须可由有权限的人复核；连接可能造成伤害的硬件时，还需要可观察、可停止和断连安全状态。</description>
      <ve:id>ve-0ab6c6e1c9c049b58de5a2125c79978d</ve:id>
      <ve:source url="https://www.anthropic.com/news/2026-usage-policy-update">Anthropic News</ve:source>
    </item>
    <item>
      <title>生产级 Agent 如何控制工具与上下文？Postman 的架构经验与企业实施清单</title>
      <link>https://vectorepoch.com/intelligence/articles/postman-production-agent-tool-context-governance/</link>
      <guid>https://vectorepoch.com/intelligence/articles/postman-production-agent-tool-context-governance/</guid>
      <description>Postman 与 AWS 公开了 Agent Mode 面向全球开发者运行时的工程经验：真正限制生产 Agent 的不只是模型能力，而是工具膨胀、界面状态耦合、上下文噪声、写操作审批和推理成本。对企业而言，可复制的方法是按任务缩小工具集、用结构化查询替代大量单用途读取工具，并为每类实体构建专用上下文。</description>
      <ve:id>ve-0fdd4310c11b47b0baecab67c8950609</ve:id>
      <ve:source url="https://aws.amazon.com/blogs/machine-learning/how-postman-runs-agent-mode-for-40-million-developers-on-amazon-bedrock/">AWS Machine Learning Blog</ve:source>
    </item>
    <item>
      <title>向量纪元 AI 产业与企业智能化日报｜2026-10-10</title>
      <link>https://vectorepoch.com/intelligence/daily/2026-10-10/</link>
      <guid>https://vectorepoch.com/intelligence/daily/2026-10-10/</guid>
      <description>本期聚焦生产级 Agent 的工具与上下文治理、工业机器人仿真资产验证、企业数据联邦与资源成本优化。事实均来自已访问的官方公开页面；厂商案例与效果数据明确归因，编辑判断和企业建议单独呈现。</description>
      <ve:id>ve-72cdb18348a04d0a992c85914c3411db</ve:id>
      <ve:source url="https://aws.amazon.com/blogs/machine-learning/how-postman-runs-agent-mode-for-40-million-developers-on-amazon-bedrock/">AWS Machine Learning Blog</ve:source>
      <ve:source url="https://www.databricks.com/blog/how-california-built-its-behavioral-health-public-county-profile-databricks">Databricks Blog</ve:source>
      <ve:source url="https://www.databricks.com/blog/solving-defense-supply-chain-visibility-through-governed-data-sharing">Databricks Blog</ve:source>
      <ve:source url="https://www.databricks.com/blog/announcing-workday-data-connect-federation-unity-catalog">Databricks Blog</ve:source>
      <ve:source url="https://www.databricks.com/blog/lakebase-and-agentic-sdlc-branching-databases-coding-agents">Databricks Blog</ve:source>
      <ve:source url="https://www.databricks.com/blog/introducing-funke-native-hl7v2-parsing-databricks">Databricks Blog</ve:source>
      <ve:source url="https://developer.nvidia.com/blog/5-steps-to-create-simready-assets-for-robotics-with-frontier-ai-models/">NVIDIA Technical Blog</ve:source>
      <ve:source url="https://www.databricks.com/blog/load-terabytes-data-minutes-lakebase-postgres">Databricks Blog</ve:source>
      <ve:source url="https://engineering.salesforce.com/apache-spark-resource-optimization-lessons-from-eight-million-jobs-a-day/">Salesforce Engineering</ve:source>
      <ve:source url="https://www.databricks.com/blog/how-build-governed-enterprise-apps-databricks-replit-and-lakebase">Databricks Blog</ve:source>
      <ve:source url="https://www.anthropic.com/news/2026-usage-policy-update">Anthropic News</ve:source>
      <ve:source url="https://developer.nvidia.com/blog/doca-gpunetio-gda-ki-unified-gpu-networking/">NVIDIA Technical Blog</ve:source>
    </item>
    <item>
      <title>Spark 资源优化为什么不能只看成功率？从执行遥测到分阶段验证</title>
      <link>https://vectorepoch.com/intelligence/articles/spark-rightsizing-execution-telemetry/</link>
      <guid>https://vectorepoch.com/intelligence/articles/spark-rightsizing-execution-telemetry/</guid>
      <description>Salesforce Engineering 公开了 Data 360 每日约 800 万次 Spark 工作负载的资源优化方法：在统一提交层采集执行遥测，用完整、可比的运行记录区分闲置资源与可靠性问题，再对单一变量进行分阶段生产验证。该案例说明，成功完成只能证明配置可运行，不能证明资源有效。</description>
      <ve:id>ve-b8a5031a06754f10be18b2cb42c8ad8d</ve:id>
      <ve:source url="https://engineering.salesforce.com/apache-spark-resource-optimization-lessons-from-eight-million-jobs-a-day/">Salesforce Engineering</ve:source>
    </item>
    <item>
      <title>机器人 CAD 如何变成可验证仿真资产？SimReady 五步流程与安全边界</title>
      <link>https://vectorepoch.com/intelligence/articles/simready-robotics-cad-validation-workflow/</link>
      <guid>https://vectorepoch.com/intelligence/articles/simready-robotics-cad-validation-workflow/</guid>
      <description>NVIDIA 以 ABB YuMi 机器人为例，把 CAD 到仿真的工作拆成 STEP 转 OpenUSD、外观匹配、物理配置、SimReady 校验和拾放任务五步。文章的核心价值不是让模型自动写代码，而是把参考资料、物理假设、验证项和运行结果放进可审查流程。</description>
      <ve:id>ve-dd13ed2b6b8a46b291a1935ecd86a67d</ve:id>
      <ve:source url="https://developer.nvidia.com/blog/5-steps-to-create-simready-assets-for-robotics-with-frontier-ai-models/">NVIDIA Technical Blog</ve:source>
    </item>
    <item>
      <title>端侧多模态检索正在成熟：工厂知识库不一定全部上传云端</title>
      <link>https://vectorepoch.com/intelligence/articles/embeddinggemma-2-on-device-multimodal-retrieval/</link>
      <guid>https://vectorepoch.com/intelligence/articles/embeddinggemma-2-on-device-multimodal-retrieval/</guid>
      <description>Google DeepMind 发布 EmbeddingGemma 2，面向文本、代码、图像、视频和音频的统一向量表示，并强调端侧运行、隐私和离线检索。它为工厂本地资料搜索提供了新的技术选项，但公开基准不等于工厂现场可直接使用。</description>
      <ve:id>ve-1503e45543ca4aa18016ffeaf7fa41f5</ve:id>
      <ve:source url="https://blog.google/innovation-and-ai/technology/developers-tools/embeddinggemma-2/">Google Blog</ve:source>
    </item>
    <item>
      <title>AI Agent 进入工业仿真：先在可验证环境里执行，再进入真实现场</title>
      <link>https://vectorepoch.com/intelligence/articles/ai-agents-enter-industrial-simulation/</link>
      <guid>https://vectorepoch.com/intelligence/articles/ai-agents-enter-industrial-simulation/</guid>
      <description>NVIDIA 展示了前沿 AI Agent 与 Omniverse 仿真能力结合的多个案例。对制造企业来说，最值得关注的不是自动生成场景本身，而是 Agent 能否在数字孪生和仿真环境中完成有指标、有人工复核的迭代任务。</description>
      <ve:id>ve-20793f770ad547ef9650b0eb26783d4c</ve:id>
      <ve:source url="https://blogs.nvidia.com/blog/developers-simulation-frontier-ai-agents/">NVIDIA Blog</ve:source>
    </item>
    <item>
      <title>边缘端决策模型走向多模态：低时延有价值，但必须用现场数据验证</title>
      <link>https://vectorepoch.com/intelligence/articles/edge-multimodal-decision-models/</link>
      <guid>https://vectorepoch.com/intelligence/articles/edge-multimodal-decision-models/</guid>
      <description>Liquid AI 发布两款开放决策模型，分别支持文本加图像，以及文本加图像或音频输入，并公布了 Jetson 设备上的响应时间。它展示了边缘设备快速分类和路由的可能性，但其中一款仍处于早期研究阶段，且官方没有提供完整视觉和音频基准。</description>
      <ve:id>ve-38e91e43851e424fa1d1080c3d53a909</ve:id>
      <ve:source url="https://huggingface.co/blog/LiquidAI/open-d1">Liquid AI 团队（Hugging Face）</ve:source>
    </item>
    <item>
      <title>企业 RAG 不能绕过原系统权限：检索时仍要向事实源确认谁能看</title>
      <link>https://vectorepoch.com/intelligence/articles/rag-access-control-must-follow-source-permissions/</link>
      <guid>https://vectorepoch.com/intelligence/articles/rag-access-control-must-follow-source-permissions/</guid>
      <description>AWS 讨论了 SharePoint、Google Drive 和 Confluence 等知识源进入企业 RAG 后的权限问题，并提出在查询时向权威来源实时确认文档级权限。对企业 Agent 来说，权限不是知识库上线前的一次配置，而是每次检索都要执行的业务规则。</description>
      <ve:id>ve-5096d77b276941c9a2c3b55473cf2e47</ve:id>
      <ve:source url="https://aws.amazon.com/blogs/machine-learning/rethinking-access-control-for-rag-with-amazon-quick-and-amazon-bedrock/">AWS Machine Learning Blog</ve:source>
    </item>
    <item>
      <title>向量纪元 AI 与制造业智能化洞察｜2026-10-09</title>
      <link>https://vectorepoch.com/intelligence/daily/2026-10-09/</link>
      <guid>https://vectorepoch.com/intelligence/daily/2026-10-09/</guid>
      <description>本期按行业日报结构汇总大模型厂商、Agent、FDE、GEO、制造业 AI 与基础设施治理动态。来源日期与本期收录日期分别呈现；未核实的发布时间留空，厂商声明、向量纪元判断与企业行动建议分别标明。</description>
      <ve:id>ve-6594e3bc29454a87a7fec1dfe3bbbec5</ve:id>
      <ve:source url="https://blogs.nvidia.com/blog/developers-simulation-frontier-ai-agents/">NVIDIA Blog</ve:source>
      <ve:source url="https://aws.amazon.com/blogs/machine-learning/pay-per-inference-for-ai-agents-how-blockrun-and-incarna-use-amazon-bedrock-agentcore-payments/">AWS Machine Learning Blog</ve:source>
      <ve:source url="https://developer.nvidia.com/blog/building-reliable-data-analytics-agents-lessons-from-the-kdd-cup/">NVIDIA Technical Blog</ve:source>
      <ve:source url="https://aws.amazon.com/blogs/machine-learning/share-gpu-clusters-across-teams-with-isolation-and-fairness-using-amazon-sagemaker-hyperpod/">AWS Machine Learning Blog</ve:source>
      <ve:source url="https://aws.amazon.com/blogs/machine-learning/rethinking-access-control-for-rag-with-amazon-quick-and-amazon-bedrock/">AWS Machine Learning Blog</ve:source>
      <ve:source url="https://developer.nvidia.com/blog/the-machines-that-make-the-machines/">NVIDIA Technical Blog</ve:source>
      <ve:source url="https://aws.amazon.com/blogs/machine-learning/beyond-hours-saved-building-the-business-case-for-agentic-automation/">AWS Machine Learning Blog</ve:source>
      <ve:source url="https://blog.google/innovation-and-ai/technology/developers-tools/embeddinggemma-2/">Google Blog</ve:source>
      <ve:source url="https://aws.amazon.com/blogs/machine-learning/introducing-claude-haiku-5-5-on-aws/">AWS Machine Learning Blog</ve:source>
      <ve:source url="https://developers.google.com/search/blog/2026/10/seo-learning-paths">Google Search Central Blog</ve:source>
      <ve:source url="https://huggingface.co/blog/LiquidAI/open-d1">Liquid AI 团队（Hugging Face）</ve:source>
    </item>
    <item>
      <title>Agent 开始按次购买推理服务：预算、授权和审计要成为基础能力</title>
      <link>https://vectorepoch.com/intelligence/articles/pay-per-inference-agent-budget-boundaries/</link>
      <guid>https://vectorepoch.com/intelligence/articles/pay-per-inference-agent-budget-boundaries/</guid>
      <description>AWS 展示了 Agent 按单次推理购买外部服务的支付方式，并由基础设施执行消费上限。机器之间能够自动购买能力后，企业需要同步建立额度、审批、幂等和账单审计，而不是只关注调用是否成功。</description>
      <ve:id>ve-fef5ceacecda45b38a066fe9ae47aa65</ve:id>
      <ve:source url="https://aws.amazon.com/blogs/machine-learning/pay-per-inference-for-ai-agents-how-blockrun-and-incarna-use-amazon-bedrock-agentcore-payments/">AWS Machine Learning Blog</ve:source>
    </item>
  </channel>
</rss>