{"id":403,"date":"2026-05-14T07:26:50","date_gmt":"2026-05-13T23:26:50","guid":{"rendered":"http:\/\/www.faiyi.com\/?p=403"},"modified":"2026-05-14T07:26:50","modified_gmt":"2026-05-13T23:26:50","slug":"ai%e5%8a%a8%e6%80%81%e6%af%8f%e6%97%a5%e7%ae%80%e6%8a%a5-2026-05-14-2","status":"publish","type":"post","link":"http:\/\/www.faiyi.com\/?p=403","title":{"rendered":"AI\u52a8\u6001\u6bcf\u65e5\u7b80\u62a5 2026-05-14"},"content":{"rendered":"

\u65e5\u671f\uff1a2026-05-14<\/p>\n

\u672c\u671f\u805a\u7126\uff1a\u91cd\u70b9\u5173\u6ce8\u6a21\u578b\u53d1\u5e03\u4e0e release notes\u3001\u5b98\u65b9 engineering blog\u3001AI coding \/ agent \/ SRE\u3001\u8bc4\u6d4b\u699c\u5355\u53d8\u5316\u3001\u5f00\u53d1\u8005\u5b9e\u8df5\u535a\u5ba2\u3001\u6846\u67b6\u751f\u6001\u3001\u5f00\u6e90\u6a21\u578b\u4e0e\u771f\u5b9e\u7528\u6237\u89c6\u89d2\uff1b\u5f53 HN\u3001Reddit\u3001Hugging Face \u7b49\u793e\u533a\u6e90\u53ef\u8bbf\u95ee\u65f6\u4f18\u5148\u7eb3\u5165\u3002<\/p>\n

\n
    \n
  1. \n

    Notion just turned its workspace into a hub for AI agents<\/strong>\uff08TechCrunch AI\uff09<\/p>\n

    \u4e2d\u6587\u6458\u8981\uff1a<\/strong>Notion \u5728\u5468\u4e09\u7684\u4ea7\u54c1\u53d1\u5e03\u4f1a\u4e0a\u63a8\u51fa\u5168\u65b0\u5f00\u53d1\u8005\u5e73\u53f0\uff0c\u5c06\u81ea\u8eab\u4ece\u534f\u4f5c\u7b14\u8bb0\u5e94\u7528\u8f6c\u578b\u4e3a AI Agent \u534f\u4f5c\u4e2d\u67a2\u3002\u65b0\u5e73\u53f0\u901a\u8fc7 Notion CLI \u652f\u6301\u5f00\u53d1\u8005\u90e8\u7f72\u81ea\u5b9a\u4e49\u4ee3\u7801\uff08Workers\uff09\uff0c\u8fde\u63a5\u5916\u90e8 Agent\uff0c\u5e76\u6784\u5efa\u53ef\u8de8\u6570\u636e\u5e93\u62c9\u53d6\u6570\u636e\u7684\u591a\u6b65\u9aa4\u81ea\u52a8\u5316\u5de5\u4f5c\u6d41\u3002CEO Ivan Zhao \u627f\u8ba4 Notion \u5386\u53f2\u4e0a\u5e76\u975e\u5f00\u53d1\u8005\u53cb\u597d\u5e73\u53f0\uff0c\u4f46\u6b63\u79ef\u6781\u6539\u53d8\u3002\u8be5 orchestration \u5c42\u4f7f Notion \u6210\u4e3a\u53ef\u7f16\u7a0b\u5e73\u53f0\uff0c\u4e0e workflow automation \u5de5\u5177\u76f4\u63a5\u7ade\u4e89\uff0c\u7784\u51c6\u4f01\u4e1a\u77e5\u8bc6\u5de5\u4f5c\u81ea\u52a8\u5316\u4e0e\u5185\u90e8 AI \u7cfb\u7edf\u5efa\u8bbe\u9700\u6c42\u3002<\/p>\n

    English Summary:<\/strong> Notion unveiled a new developer platform in a live-streamed announcement, transforming itself from a collaborative note-taking app into a hub for AI agent collaboration. The platform, accessible via Notion CLI, allows developers to deploy custom code through Workers, connect external agents, and build automated multi-step workflows that pull data across databases. CEO Ivan Zhao acknowledged Notion's historically limited developer focus but emphasized the shift.<\/p>\n

    \u539f\u6587\u94fe\u63a5<\/a><\/p>\n<\/li>\n

  2. \n

    Anthropic Launches Claude Platform on AWS<\/strong>\uff08InfoQ AI\/ML\uff09<\/p>\n

    \u4e2d\u6587\u6458\u8981\uff1a<\/strong>Anthropic \u5ba3\u5e03 Claude Platform \u5728 AWS \u4e0a\u6b63\u5f0f GA\uff0c\u4e3a AWS \u5ba2\u6237\u63d0\u4f9b\u76f4\u63a5\u4f7f\u7528 Anthropic \u539f\u751f Claude \u5e73\u53f0\u7684\u65b0\u90e8\u7f72\u9009\u9879\u3002\u8be5\u96c6\u6210\u5141\u8bb8\u5ba2\u6237\u4f7f\u7528 AWS \u8ba4\u8bc1\u3001\u8ba1\u8d39\u548c\u76d1\u63a7\u670d\u52a1\u8bbf\u95ee Claude\uff0c\u7b80\u5316\u4f01\u4e1a\u7ea7 AI \u670d\u52a1\u7684\u91c7\u8d2d\u4e0e\u6cbb\u7406\u6d41\u7a0b\u3002\u6b64\u4e3e\u6807\u5fd7\u7740 Anthropic \u6df1\u5316\u4e0e\u4e91\u5382\u5546\u7684\u5408\u4f5c\uff0c\u4e3a\u4f01\u4e1a\u7528\u6237\u63d0\u4f9b\u66f4\u65e0\u7f1d\u7684\u6a21\u578b\u63a5\u5165\u4f53\u9a8c\uff0c\u540c\u65f6\u501f\u52a9 AWS \u7684\u4f01\u4e1a\u57fa\u7840\u8bbe\u65bd\u4f18\u52bf\u6269\u5927 Claude \u7684\u5e02\u573a\u8986\u76d6\u3002<\/p>\n

    English Summary:<\/strong> Anthropic announced the general availability of Claude Platform on AWS, offering AWS customers direct access to Anthropic's native Claude platform using AWS authentication, billing, and monitoring services. This integration simplifies procurement and governance for enterprise AI services, marking Anthropic's deepening partnership with cloud providers. It provides seamless model access for enterprise users while leveraging AWS's infrastructure advantages to expand Claude's market reach.<\/p>\n

    \u539f\u6587\u94fe\u63a5<\/a><\/p>\n<\/li>\n

  3. \n

    Build financial document processing with Pulse AI and Amazon Bedrock<\/strong>\uff08AWS ML Blog\uff09<\/p>\n

    \u4e2d\u6587\u6458\u8981\uff1a<\/strong>AWS \u673a\u5668\u5b66\u4e60\u535a\u5ba2\u53d1\u5e03\u4e0e Pulse AI \u5408\u4f5c\u7684\u91d1\u878d\u6587\u6863\u5904\u7406\u65b9\u6848\uff0c\u5c55\u793a\u5982\u4f55\u6784\u5efa\u6587\u6863\u63d0\u53d6\u4e0e\u6a21\u578b\u5fae\u8c03\u6d41\u6c34\u7ebf\u4ee5\u5e94\u5bf9\u590d\u6742\u91d1\u878d\u6587\u6863\u5904\u7406\u6311\u6218\u3002Pulse \u5c06\u975e\u7ed3\u6784\u5316\u6587\u6863\u8f6c\u6362\u4e3a\u7ed3\u6784\u5316\u3001schema \u5bf9\u9f50\u7684\u8f93\u51fa\uff0c\u5e76\u53ef\u76f4\u63a5\u5bfc\u51fa\u4e3a\u517c\u5bb9 Amazon Bedrock \u5fae\u8c03\u8981\u6c42\u7684 JSONL \u6570\u636e\u96c6\u3002\u8be5\u65b9\u6848\u8ba9 Pulse \u8d1f\u8d23\u63d0\u53d6\u4e0e\u6570\u636e\u8d28\u91cf\u7684\u91cd\u6d3b\uff0c\u540c\u65f6\u7b80\u5316\u5728 Bedrock \u4e0a\u8fdb\u884c\u6587\u672c\u548c\u89c6\u89c9\u6a21\u578b\u5b9a\u5236\u6240\u9700\u7684\u8bad\u7ec3\u6570\u636e\u751f\u6210\u6d41\u7a0b\uff0c\u652f\u6301\u8fed\u4ee3\u5fae\u8c03\u4ee5\u5b9e\u73b0\u6301\u7eed\u6539\u8fdb\u3002<\/p>\n

    English Summary:<\/strong> AWS Machine Learning Blog published a solution for financial document processing using Pulse AI and Amazon Bedrock, demonstrating how to build extraction and model fine-tuning pipelines for complex financial documents. Pulse converts unstructured documents into structured, schema-aligned outputs exportable as JSONL datasets compatible with Amazon Bedrock's fine-tuning requirements.<\/p>\n

    \u539f\u6587\u94fe\u63a5<\/a><\/p>\n<\/li>\n

  4. \n

    TypeScript, C# and Turbo Pascal with Anders Hejlsberg<\/strong>\uff08Pragmatic Engineer\uff09<\/p>\n

    \u4e2d\u6587\u6458\u8981\uff1a<\/strong>TypeScript\u3001C# \u4e0e Turbo Pascal \u7684\u7f14\u9020\u8005 Anders Hejlsberg \u5728 Pragmatic Engineer \u64ad\u5ba2\u4e2d\u56de\u987e\u5176\u7f16\u7a0b\u8bed\u8a00\u8bbe\u8ba1\u751f\u6daf\uff0c\u5e76\u5206\u4eab\u5bf9 AI \u5982\u4f55\u6539\u53d8\u8f6f\u4ef6\u5de5\u7a0b\u672a\u6765\u7684\u89c1\u89e3\u3002\u4ed6\u6307\u51fa AI \u5728\u7279\u5b9a\u8bed\u8a00\u4e0a\u7684\u8868\u73b0\u53d6\u51b3\u4e8e\u8bad\u7ec3\u6570\u636e\u91cf\uff0cTypeScript \u548c Python \u8868\u73b0\u4f18\u5f02\u6b63\u56e0\u4e92\u8054\u7f51\u4e0a\u6709\u5927\u91cf\u6837\u672c\uff1b\u4f46 AI \u76ee\u524d\u5728\u7f16\u5199\u7f16\u8bd1\u5668\u65b9\u9762\u4ecd\u6709\u5c40\u9650\uff0c\u96be\u4ee5\u628a\u63e1\u7c7b\u578b\u3001\u7b26\u53f7\u3001\u7ed1\u5b9a\u4e0e\u89e3\u6790\u4e4b\u95f4\u7684\u5168\u5c40\u5173\u7cfb\u3002Hejlsberg \u9884\u6d4b\u672a\u6765\u5f00\u53d1\u8005\u5c06\u66f4\u50cf\u9879\u76ee\u7ecf\u7406\uff0c\u7ba1\u7406\u5927\u91cf\u751f\u6210\u4ee3\u7801\u7684 Agent\uff0c\u800c\u4ee3\u7801\u5ba1\u67e5\u53ef\u80fd\u6210\u4e3a\u65b0\u7684\u6838\u5fc3\u6280\u827a\u3002<\/p>\n

    English Summary:<\/strong> Anders Hejlsberg, creator of TypeScript, C#, and Turbo Pascal, reflected on his language design career and shared insights on AI's impact on software engineering in the Pragmatic Engineer podcast. He noted AI performance in specific languages depends on training data volume\u2014TypeScript and Python excel due to abundant online samples\u2014while AI remains limited in writing compilers, struggling to see the big picture across types, symbols, binding, and parsing.<\/p>\n

    \u539f\u6587\u94fe\u63a5<\/a><\/p>\n<\/li>\n

  5. \n

    Quoting Boris Mann<\/strong>\uff08Simon Willison\uff09<\/p>\n

    \u4e2d\u6587\u6458\u8981\uff1a<\/strong>\u5f00\u53d1\u8005 Boris Mann \u5728 Simon Willison \u535a\u5ba2\u7684\u5f15\u7528\u4e2d\u6307\u51fa\uff0c"11 \u4e2a AI Agent" \u8fd9\u4e00\u8868\u8ff0\u4f5c\u4e3a\u77ed\u8bed\u6beb\u65e0\u610f\u4e49\uff0c\u7c7b\u4f3c\u4e8e\u8bf4"\u6211\u6709 11 \u4e2a\u7535\u5b50\u8868\u683c"\u6216"\u6211\u6709 11 \u4e2a\u6d4f\u89c8\u5668\u6807\u7b7e\u9875"\u6765\u5b8c\u6210\u5de5\u4f5c\u3002\u8be5\u89c2\u70b9\u5f15\u53d1\u5bf9\u5f53\u524d AI Agent \u6982\u5ff5\u6cdb\u5316\u4e0e\u8425\u9500\u8bdd\u672f\u7684\u53cd\u601d\uff0c\u5f3a\u8c03 Agent \u6570\u91cf\u672c\u8eab\u5e76\u4e0d\u80fd\u8bf4\u660e\u5de5\u4f5c\u6d41\u7684\u5b9e\u9645\u4ef7\u503c\u6216\u590d\u6742\u6027\uff0c\u547c\u5401\u884c\u4e1a\u5173\u6ce8 Agent \u7684\u5b9e\u9645\u80fd\u529b\u4e0e\u534f\u4f5c\u8d28\u91cf\u800c\u975e\u7b80\u5355\u8ba1\u6570\u3002<\/p>\n

    English Summary:<\/strong> Developer Boris Mann, quoted on Simon Willison's blog, argues that "11 AI agents" is meaningless as a phrase\u2014akin to saying "I have 11 spreadsheets" or "I have 11 browser tabs" to describe one's work. This observation sparks reflection on the current overgeneralization and marketing hype around AI agents, emphasizing that agent count alone doesn't indicate actual workflow value or complexity, and calling for the industry to focus on agent capabilities and collaboration quality rather than simple enumeration.<\/p>\n

    \u539f\u6587\u94fe\u63a5<\/a><\/p>\n<\/li>\n

  6. \n

    Building a safe, effective sandbox to enable Codex on Windows<\/strong>\uff08OpenAI News\uff09<\/p>\n

    \u4e2d\u6587\u6458\u8981\uff1a<\/strong>OpenAI \u53d1\u5e03\u6280\u672f\u535a\u5ba2\u8be6\u7ec6\u9610\u8ff0\u5982\u4f55\u5728 Windows \u5e73\u53f0\u4e3a Codex \u6784\u5efa\u5b89\u5168\u6c99\u7bb1\u3002\u7531\u4e8e Windows \u7f3a\u4e4f\u7c7b\u4f3c macOS Seatbelt \u6216 Linux seccomp \u7684\u539f\u751f\u9694\u79bb\u673a\u5236\uff0c\u56e2\u961f\u8bc4\u4f30\u4e86 AppContainer\u3001Windows Sandbox \u548c MIC \u7b49\u65b9\u6848\u540e\u5747\u653e\u5f03\uff0c\u6700\u7ec8\u81ea\u4e3b\u8bbe\u8ba1\u4e86\u4e00\u5957\u57fa\u4e8e\u5408\u6210 SID\uff08sandbox-write\uff09\u548c\u5199\u9650\u5236\u4ee4\u724c\uff08write-restricted token\uff09\u7684\u6743\u9650\u7cfb\u7edf\uff0c\u5b9e\u73b0\u65e0\u9700\u7ba1\u7406\u5458\u6743\u9650\u5373\u53ef\u9650\u5236\u6587\u4ef6\u5199\u5165\u8303\u56f4\u3002\u7f51\u7edc\u9694\u79bb\u65b9\u9762\uff0c\u65e9\u671f\u539f\u578b\u4ec5\u901a\u8fc7\u73af\u5883\u53d8\u91cf\u548c PATH \u52ab\u6301\u8fdb\u884c\u8f6f\u6027\u963b\u65ad\uff0c\u4f46\u5b58\u5728\u88ab\u7ed5\u8fc7\u7684\u98ce\u9669\uff1b\u6700\u7ec8\u7248\u672c\u5f15\u5165\u9632\u706b\u5899\u89c4\u5219\uff0c\u521b\u5efa CodexSandboxOffline\/Online \u4e24\u4e2a\u4e13\u7528\u672c\u5730\u7528\u6237\uff0c\u5b9e\u73b0\u5f3a\u5236\u7684\u7f51\u7edc\u8bbf\u95ee\u63a7\u5236\u3002\u6574\u4e2a\u67b6\u6784\u5305\u542b codex.exe\u3001codex-windows-sandbox-setup.exe\uff08\u63d0\u6743\u8bbe\u7f6e\uff09\u548c codex-command-runner.exe\uff08\u547d\u4ee4\u6267\u884c\u5668\uff09\u4e09\u4e2a\u6838\u5fc3\u7ec4\u4ef6\uff0c\u5728\u5b89\u5168\u6027\u4e0e\u5f00\u53d1\u8005\u5de5\u4f5c\u6d41\u517c\u5bb9\u6027\u4e4b\u95f4\u53d6\u5f97\u5e73\u8861\u3002<\/p>\n

    English Summary:<\/strong> OpenAI's engineering blog details building a secure sandbox for Codex on Windows. Without native OS isolation primitives like macOS Seatbelt or Linux seccomp, the team evaluated and rejected AppContainer, Windows Sandbox, and MIC. They designed a custom solution using synthetic SIDs (sandbox-write) and write-restricted tokens to limit filesystem writes without requiring admin privileges. For network isolation, the initial prototype used environment variable hijacking but was advisory-only; the final implementation uses Windows Firewall rules applied to dedicated local users (CodexSandboxOffline\/Online). The architecture comprises three binaries: codex.exe, an elevated setup utility, and a command runner, balancing security with practical developer workflows.<\/p>\n

    \u539f\u6587\u94fe\u63a5<\/a><\/p>\n<\/li>\n

  7. \n

    CSP Allow-list Experiment<\/strong>\uff08Simon Willison\uff09<\/p>\n

    \u4e2d\u6587\u6458\u8981\uff1a<\/strong>Simon Willison \u53d1\u5e03\u5b9e\u9a8c\u5de5\u5177 CSP Allow-list\uff0c\u5c55\u793a\u5982\u4f55\u5728\u53d7\u5185\u5bb9\u5b89\u5168\u7b56\u7565\uff08CSP\uff09\u4fdd\u62a4\u7684\u6c99\u7bb1 iframe \u4e2d\u52a0\u8f7d\u5e94\u7528\uff0c\u5e76\u901a\u8fc7\u81ea\u5b9a\u4e49 fetch() \u62e6\u622a CSP \u8fdd\u89c4\u9519\u8bef\uff0c\u5c06\u4fe1\u606f\u4f20\u9012\u81f3\u7236\u7a97\u53e3\uff0c\u4ece\u800c\u63d0\u793a\u7528\u6237\u5c06\u7279\u5b9a\u57df\u540d\u52a0\u5165\u767d\u540d\u5355\u5e76\u5237\u65b0\u9875\u9762\u3002\u8be5\u5b9e\u9a8c\u57fa\u4e8e\u6b64\u524d\u5173\u4e8e\u6d4b\u8bd5 CSP iframe \u9003\u9038\u7684\u7814\u7a76\uff0c\u4f7f\u7528 GPT-5.5\uff08xhigh \u6a21\u5f0f\uff09\u5728 Codex \u684c\u9762\u5e94\u7528\u4e2d\u5f00\u53d1\u5b8c\u6210\u3002\u5de5\u5177\u6f14\u793a\u4e86\u4e00\u79cd\u5728\u4e25\u683c CSP \u73af\u5883\u4e0b\u5b9e\u73b0\u52a8\u6001\u57df\u540d\u6388\u6743\u7684\u6280\u672f\u8def\u5f84\uff0c\u5bf9\u9700\u8981\u5728\u53d7\u9650\u73af\u5883\u4e2d\u8fd0\u884c\u7b2c\u4e09\u65b9\u4ee3\u7801\u7684\u5f00\u53d1\u8005\u5177\u6709\u53c2\u8003\u4ef7\u503c\u3002<\/p>\n

    English Summary:<\/strong> Simon Willison released the CSP Allow-list Experiment, demonstrating how to load an app inside a CSP-protected sandboxed iframe with a custom fetch() that intercepts CSP errors and passes them to the parent window. This allows prompting users to add domains to an allow-list and refresh. Built with GPT-5.5 xhigh in the Codex desktop app, the tool showcases a technical approach to dynamic domain authorization under strict CSP constraints, relevant for developers running third-party code in restricted environments.<\/p>\n

    \u539f\u6587\u94fe\u63a5<\/a><\/p>\n<\/li>\n

  8. \n

    [AINews] The End of Finetuning<\/strong>\uff08Latent Space\uff09<\/p>\n

    \u4e2d\u6587\u6458\u8981\uff1a<\/strong>Latent Space \u53d1\u5e03\u8bc4\u8bba\u6587\u7ae0\u63a2\u8ba8\u5fae\u8c03\uff08finetuning\uff09\u7684\u7ec8\u7ed3\u8d8b\u52bf\u3002\u6587\u7ae0\u6307\u51fa OpenAI \u8fd1\u671f\u5f03\u7528\u5176\u5fae\u8c03 API \u662f\u8fd9\u4e00\u8d8b\u52bf\u7684\u6700\u65b0\u6807\u5fd7\u2014\u2014\u5c3d\u7ba1 OpenAI \u66fe\u662f\u5927\u578b\u5b9e\u9a8c\u5ba4\u4e2d\u5fae\u8c03\u652f\u6301\u7684\u4ee3\u8868\uff0c\u5e76\u957f\u671f\u5ba3\u4f20"\u4ee5 o1 \u7684\u6027\u80fd\u83b7\u5f97 4o \u7684\u4ef7\u683c"\u8fd9\u4e00\u7406\u5ff5\u3002\u4f5c\u8005\u8ba4\u4e3a\uff0c\u5373\u4f7f\u6ca1\u6709\u6781\u7aef\u7684 GPU \u4f9b\u5e94\u7d27\u5f20\uff0cAI \u5de5\u7a0b\u884c\u4e1a\u7684\u4e3b\u6d41\u8d8b\u52bf\u4e5f\u5728\u5411\u63d0\u793a\u5de5\u7a0b\uff08prompt engineering\uff09\u548c\u4e0a\u4e0b\u6587\u5b66\u4e60\u8f6c\u79fb\uff0cFast.ai \u7684 Jeremy Howard \u65e9\u5728 2023 \u5e74\u5c31\u5df2\u9884\u8a00\u8fd9\u4e00\u70b9\u3002\u4e0d\u8fc7\u6587\u7ae0\u4e5f\u5f3a\u8c03\uff0c\u8fd9\u5e76\u4e0d\u610f\u5473\u7740\u5fae\u8c03\u5b8c\u5168\u6d88\u5931\uff1aCursor\u3001Cognition \u7b49\u9876\u7ea7\u5e94\u7528\u516c\u53f8\u53cd\u800c\u589e\u52a0\u4e86\u5f00\u653e\u6a21\u578b\u7684 RLFT\uff08\u5f3a\u5316\u5b66\u4e60\u5fae\u8c03\uff09\u4f7f\u7528\uff0c\u5f00\u6e90\u6a21\u578b\u5fae\u8c03\u5728\u5b9a\u5236 ASIC \u63a8\u7406\u65b9\u6848\u4e2d\u4e5f\u53ef\u80fd\u626e\u6f14\u5173\u952e\u89d2\u8272\u3002<\/p>\n

    English Summary:<\/strong> Latent Space published an op-ed on the declining role of finetuning in AI engineering. OpenAI's recent deprecation of its finetuning APIs signals a broader shift, despite years of promoting "o1 performance at 4o prices." The author argues the industry was already trending toward prompt engineering and in-context learning, as Jeremy Howard predicted in 2023. However, the "end" of finetuning for most doesn't mean its total disappearance: top-tier companies like Cursor and Cognition have increased RLFT usage on open models, and open-model finetuning may remain central to custom ASIC inference strategies.<\/p>\n

    \u539f\u6587\u94fe\u63a5<\/a><\/p>\n<\/li>\n

  9. \n

    Announcements Introducing Claude for Small Business<\/strong>\uff08Anthropic News\uff09<\/p>\n

    \u4e2d\u6587\u6458\u8981\uff1a<\/strong>Anthropic \u63a8\u51fa Claude for Small Business\uff0c\u9762\u5411\u5c0f\u578b\u4f01\u4e1a\u63d0\u4f9b\u8fde\u63a5\u5668\u4e0e\u5f00\u7bb1\u5373\u7528\u7684\u5de5\u4f5c\u6d41\uff0c\u5c06 Claude \u96c6\u6210\u81f3 QuickBooks\u3001PayPal\u3001HubSpot\u3001Canva\u3001Docusign\u3001Google Workspace \u548c Microsoft 365 \u7b49\u5e38\u7528\u5de5\u5177\u3002\u8be5\u65b9\u6848\u5305\u542b 15 \u4e2a\u9884\u8bbe\u7684\u667a\u80fd\u4f53\u5de5\u4f5c\u6d41\uff08\u6db5\u76d6\u8d22\u52a1\u3001\u8fd0\u8425\u3001\u9500\u552e\u3001\u8425\u9500\u3001\u4eba\u529b\u8d44\u6e90\u548c\u5ba2\u670d\uff09\u4ee5\u53ca 15 \u9879\u9488\u5bf9\u5e38\u89c1\u91cd\u590d\u4efb\u52a1\u4f18\u5316\u7684\u6280\u80fd\uff0c\u5982\u5de5\u8d44\u89c4\u5212\u3001\u6708\u672b\u7ed3\u8d26\u3001\u53d1\u7968\u50ac\u6536\u3001\u8425\u9500\u6d3b\u52a8\u6267\u884c\u7b49\u3002\u6240\u6709\u64cd\u4f5c\u9700\u7ecf\u7528\u6237\u5ba1\u6279\u540e\u65b9\u53ef\u6267\u884c\uff0c\u5e76\u7ee7\u627f\u539f\u6709\u7cfb\u7edf\u7684\u6743\u9650\u8bbe\u7f6e\u3002Anthropic \u8fd8\u4e0e PayPal \u5408\u4f5c\u63a8\u51fa\u514d\u8d39\u7684 AI \u7d20\u517b\u5728\u7ebf\u8bfe\u7a0b\uff0c\u5e76\u542f\u52a8\u8986\u76d6\u7f8e\u56fd\u591a\u57ce\u5e02\u7684\u7ebf\u4e0b\u57f9\u8bad\u5de1\u5c55\uff0c\u65e8\u5728\u7f29\u5c0f\u5c0f\u578b\u4f01\u4e1a\u4e0e\u5927\u578b\u4f01\u4e1a\u5728 AI \u5e94\u7528\u4e0a\u7684\u5dee\u8ddd\u3002<\/p>\n

    English Summary:<\/strong> Anthropic launched Claude for Small Business, offering connectors and ready-to-run workflows that integrate Claude into tools like QuickBooks, PayPal, HubSpot, Canva, Docusign, Google Workspace, and Microsoft 365. The package includes 15 agentic workflows across finance, operations, sales, marketing, HR, and customer service, plus 15 skills targeting repetitive tasks such as payroll planning, month-end close, invoice chasing, and campaign execution. All actions require user approval and inherit existing permissions. Anthropic also partnered with PayPal on a free AI fluency course and launched a multi-city training tour to help close the AI adoption gap between small businesses and enterprises.<\/p>\n

    \u539f\u6587\u94fe\u63a5<\/a><\/p>\n<\/li>\n

  10. \n

    How NVIDIA engineers and researchers build with Codex<\/strong>\uff08OpenAI News\uff09<\/p>\n

    \u4e2d\u6587\u6458\u8981\uff1a<\/strong>NVIDIA \u5de5\u7a0b\u56e2\u961f\u5206\u4eab\u4f7f\u7528 Codex\uff08\u57fa\u4e8e GPT-5.5\uff09\u7684\u5b9e\u8df5\u6848\u4f8b\u3002Codex \u5df2\u6210\u4e3a NVIDIA \u5de5\u7a0b\u5e08\u5904\u7406\u590d\u6742\u5de5\u7a0b\u4efb\u52a1\u7684\u9996\u9009\u5de5\u5177\uff0c\u5e2e\u52a9\u5185\u90e8\u5e73\u53f0\u4ece MVP \u6f14\u8fdb\u4e3a\u751f\u4ea7\u7ea7\u7cfb\u7edf\uff0c\u5e76\u5728\u6570\u5c0f\u65f6\u5185\u6784\u5efa\u51fa\u5185\u90e8\u64ad\u5ba2\u5f55\u5236\u5e94\u7528\uff08\u542b\u97f3\u89c6\u9891\u6d4b\u8bd5\uff09\uff0c\u800c\u4f20\u7edf\u91c7\u8d2d\u6d41\u7a0b\u53ef\u80fd\u9700\u8981\u6570\u5468\u3002\u7814\u7a76\u56e2\u961f\u5219\u5229\u7528 Codex \u5b9e\u73b0\u7aef\u5230\u7aef\u673a\u5668\u5b66\u4e60\u7814\u7a76\u6d41\u7a0b\u7684\u81ea\u52a8\u5316\uff1a\u4ece\u6587\u732e\u7efc\u8ff0\u3001\u5047\u8bbe\u751f\u6210\u5230\u5b9e\u9a8c\u811a\u672c\u7f16\u5199\u548c\u8fdc\u7a0b\u96c6\u7fa4\u6267\u884c\uff0c\u7814\u7a76\u6548\u7387\u63d0\u5347\u7ea6 10 \u500d\u3002\u7814\u7a76\u4eba\u5458\u8fd8\u4f7f\u7528 Codex \u8fdb\u884c\u4ee3\u7801\u8fc1\u79fb\uff0c\u5c06 Python \u4ee3\u7801\u5e93\u91cd\u5199\u4e3a Rust \u4ee5\u83b7\u5f97\u9ad8\u8fbe 20 \u500d\u7684\u6027\u80fd\u63d0\u5347\u3002\u76ee\u524d NVIDIA \u5df2\u6709\u7ea6 4 \u4e07\u540d\u5458\u5de5\u83b7\u5f97 Codex \u8bbf\u95ee\u6743\u9650\u3002<\/p>\n

    English Summary:<\/strong> NVIDIA engineers shared how they use Codex (powered by GPT-5.5) in production. Codex has become the default tool for complex engineering tasks, helping evolve internal platforms from MVP to production-ready systems and building an internal podcast recording app with audio\/video testing in hours instead of weeks. Research teams use Codex to automate end-to-end ML workflows\u2014from literature review and hypothesis generation to experiment scripting and remote execution\u2014achieving roughly 10x speed improvements. Engineers also use it for machine translation, rewriting Python codebases to Rust for up to 20x efficiency gains. Approximately 40,000 NVIDIANs currently have access to Codex.<\/p>\n

    \u539f\u6587\u94fe\u63a5<\/a><\/p>\n<\/li>\n<\/ol>\n","protected":false},"excerpt":{"rendered":"

    \u65e5\u671f\uff1a2026-05-14 \u672c\u671f\u805a\u7126\uff1a\u91cd\u70b9\u5173\u6ce8\u6a21\u578b\u53d1\u5e03\u4e0e release notes\u3001\u5b98\u65b9 engineeri […]<\/p>\n","protected":false},"author":0,"featured_media":0,"comment_status":"open","ping_status":"open","sticky":false,"template":"","format":"standard","meta":{"footnotes":""},"categories":[7],"tags":[],"class_list":["post-403","post","type-post","status-publish","format-standard","hentry","category-ai-daily"],"_links":{"self":[{"href":"http:\/\/www.faiyi.com\/index.php?rest_route=\/wp\/v2\/posts\/403","targetHints":{"allow":["GET"]}}],"collection":[{"href":"http:\/\/www.faiyi.com\/index.php?rest_route=\/wp\/v2\/posts"}],"about":[{"href":"http:\/\/www.faiyi.com\/index.php?rest_route=\/wp\/v2\/types\/post"}],"replies":[{"embeddable":true,"href":"http:\/\/www.faiyi.com\/index.php?rest_route=%2Fwp%2Fv2%2Fcomments&post=403"}],"version-history":[{"count":0,"href":"http:\/\/www.faiyi.com\/index.php?rest_route=\/wp\/v2\/posts\/403\/revisions"}],"wp:attachment":[{"href":"http:\/\/www.faiyi.com\/index.php?rest_route=%2Fwp%2Fv2%2Fmedia&parent=403"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"http:\/\/www.faiyi.com\/index.php?rest_route=%2Fwp%2Fv2%2Fcategories&post=403"},{"taxonomy":"post_tag","embeddable":true,"href":"http:\/\/www.faiyi.com\/index.php?rest_route=%2Fwp%2Fv2%2Ftags&post=403"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}