OpenAI发布AI代理Dots及办公套件拟融资300亿估值达1.4万亿美元

作者：亿邦AI 文章来源：亿邦动力

2026-09-30 12:27

作者：亿邦AI 文章来源：亿邦动力

2026-09-30 12:27

邦小白快读

全文速览

总1 Dots是什么，怎么用

1. Dots是OpenAI发布的常驻AI代理，基于GPT-6 Astra模型，拥有独立云电脑和浏览器，可接入超4000款应用，关闭聊天窗后仍持续运行，自动推进项目、监控信息、操作软件，完成后提交审核。

2. 覆盖场景包括收集客户反馈、定位修复bug、营销物料更新、科研数据分析、销售提案、内容剪辑等。

3. 目前Pro和Business Premium用户可免费获得1个Dots，后续多代理需单独计费；欧洲部分区域Pro用户暂不可用，企业可申请beta测试。

总2 安全与隐私要点

1. 后台主动研究模式只读，修改内容需校验自定义规则，修改密码等敏感操作必须人工确认。

2. 可查看Activity View工作记录并介入接管，凭证以不暴露明文密码的方式使用。

3. 企业工作区内容默认不用于模型训练，个人可自主选择；但OpenAI提醒Dots仍可能出错，重要工作需复核。

总3 配套办公新品

1. ChatGPT Space替代Library，支持团队和AI代理共享协作，内置Pages对标Google Docs和Word。

2. 幻灯片可自然语言生成，多人及AI共同编辑，后续还有表格；可配置自动化工作流同步项目信息。

总4 行业动态

1. ChatGPT周活用户12亿，OpenAI拟融资300亿美元估值1.4万亿美元。

总1 品牌营销新工具

1. Dots可学习品牌定位与创意标准，产品信息更新后同步修改上线物料，适合品牌方管理多渠道营销内容。

2. 专属Dots可承担邮件营销、客户支持等工作，企业组织身份和权限可管理，营销自动化更深入。

总2 用户行为与趋势

1. ChatGPT周活12亿，AI代理成为用户工作新入口，品牌需关注用户在AI代理生态中的交互习惯。

2. Meta Muse代理上线16天下载230万到430万，约三分之一用户绑定商业账户，显示中小商家和品牌正快速接入AI代理。

总3 价格竞争与产品策略

1. OpenAI新推每月500美元ChatGPT Pro档位，原200美元档调整限额，体现高端订阅分层策略。

2. 成本更低的GPT-6.1 Sol模型发布，降低AI功能应用门槛，品牌产品研发可考虑集成AI能力。

总1 增长市场与机会

1. 常驻AI代理正成为新赛道，Meta Muse上线16天下载量230万到430万，登顶应用商店，超1500家企业申请集成，商业应用需求爆发。

2. Muse已接入Shopify、QuickBooks、Stripe、Canva等工具，约三分之一用户绑定商业账户，中小卖家可借助AI代理处理店铺运营和支付场景。

总2 销售场景的应用

1. Dots可匹配客户需求与产品文档、识别待完成技术测试项、搭建概念验证方案、随谈判进展更新提案，适合复杂销售流程。

2. OpenAI推出Agents API计算机使用能力和插件生态，开发者可定制代理功能，形成新商业模式。

总3 风险与应对

1. Dots仍可能出错，涉及重要后果需人工复核；卖家应用AI代理时需设置审批机制和敏感操作人工确认。

2. 新Pro订阅500美元/月，使用AI代理需评估成本；欧洲部分区域用户暂不可用，需关注区域限制。

3. OpenAI拟融资300亿美元估值1.4万亿美元，显示AI行业资本热度，可跟随头部平台生态把握机会。

总1 产品改进与客户反馈

1. Dots可监测客户反馈、定位常见bug、完成小规模修复测试并提交带演示视频的代码合并请求，帮助工厂技术团队高效收集问题并迭代产品。

2. 可让Dots在科研场景随新实验数据更新分析结论与图表，用于生产流程的数据分析。

总2 数字化与电商启示

1. 专属Dots可承担采购、发票处理、客户支持、合同管理等固定岗位职责，工厂可将重复性流程交给AI代理，降低人工成本。

2. 自动化工作流可从Slack、邮件、日历同步行动项，实时更新负责人、截止时间及下一步计划，提升生产项目协同效率。

3. 成本更低的GPT-6.1 Sol模型发布，降低中小工厂采用AI技术的门槛，可探索结合业务系统。

总3 商业机会

1. 常驻AI代理市场快速增长，第三方工具集成需求旺盛（Muse超1500家企业申请），工厂可考虑为AI代理生态提供产品数据接入或开发配套工具。

总1 新技术与趋势

1. OpenAI发布常驻AI代理Dots，基于GPT-6 Astra，独立云电脑浏览器，接入4000+应用，代表AI从对话式转向主动执行式。

2. Agents API新增计算机使用能力、事件触发自动化、更深层Slack与Teams集成，服务商可基于这些能力为客户构建自动化方案。

总2 客户痛点与解决方案

1. 客户担忧AI代理出错和安全问题，OpenAI提供多层权限管控：只读模式、自定义规则、敏感操作人工确认、Activity View介入接管、凭证不暴露明文密码。

2. 数据隐私方面，企业工作区内容默认不用于训练，个人可自主选择，不用代理后台内容和私有笔记训练模型。

3. 但OpenAI提醒Dots仍可能出错，服务商需在实施中设计人工复核环节。

总3 竞争与商机

1. 常驻AI代理赛道竞争激烈，Meta Muse下载量增长迅速，服务商可关注跨平台代理接入需求。

2. OpenAI推出每月500美元Pro档位和低成本Sol模型，形成不同层级服务能力，服务商可按客户预算选择。

总1 平台生态开放

1. OpenAI扩展Agents API计算机使用能力、插件生态和事件触发自动化，并深化Slack与Teams集成，平台方可以此为参考开放API和插件市场。

2. Dots可接入超4000款第三方应用，通过网页、桌面、移动端及Slack、Teams交互，未来开放短信和语音，平台需适配多渠道代理交互。

总2 招商与合作

1. OpenAI与微软合作，将专属Dots接入Microsoft Agent 365体系，方便企业通过现有安全管理体系做权限管控和审计，平台方可借鉴企业级代理集成模式。

2. Meta Muse已接入Shopify、QuickBooks、Stripe、Canva等工具，超1500家企业申请集成，说明平台接入AI代理生态可带来新用户和商业场景。

总3 运营管理与风险

1. Dots提供Activity View工作记录、只读与人工确认机制，平台运营方需建立类似的代理行为审核和干预能力。

2. OpenAI因安全问题暂缓发布GPT-6.1 Astra，并支持放缓高级AI模型研发，平台应重视AI安全风向，避免高风险功能上线。

总1 产业新动向

1. OpenAI发布常驻AI代理Dots，标志AI助手从用户主动对话转向后台持续运行、自动操作软件，推动人机协作模式变革。

2. 常驻AI代理赛道快速拥挤：Meta Muse上线16天下载量230万到430万，另有Instinct、Manus 2.0、Grok Bot、OpenClaw等，成为AI行业核心竞争方向。

总2 新问题与政策启示

1. 大会场外有民间组织抗议，呼吁公众利益优先、终止ICE合同，反映AI伦理和社会责任争议。

2. OpenAI披露数起模型非预期行为事件，暂缓发布GPT-6.1 Astra，并公开支持放缓高级AI研发，提示监管需关注前沿模型风险。

3. Dots设置多层权限管控和数据训练选择机制，为AI代理治理提供了企业级产品设计范例。

总3 商业模式

1. OpenAI拟Pre-IPO融资300亿美元，估值1.4万亿美元，年化营收400亿美元、增长70%，暂不IPO优先AI安全，显示高风险高投入AI公司的资本路径。

2. 订阅分层：每月500美元Pro档、200美元档调整限额，Dots按额外代理和工作负载单独计费，形成基于使用量和功能等级的变现模式。

声明：快读内容全程由AI生成，请注意甄别信息。如您发现问题，请发送邮件至 run@ebrun.com 。

我是品牌商卖家工厂服务商平台商研究者帮我再读一遍。

Quick Summary

Section 1: What Dots is and how to use it

1. Dots is an always-on AI agent released by OpenAI, based on the GPT-6 Astra model. It has an independent cloud computer and browser, can connect to over 4,000 apps, continues running after the chat window is closed, automatically advances projects, monitors information, operates software, and submits results for review after completion.

2. Use cases include collecting customer feedback, locating and fixing bugs, updating marketing materials, analyzing research data, preparing sales proposals, and content editing.

3. Currently, Pro and Business Premium users can get one Dots for free; deploying additional agents requires separate billing. Pro users in some European regions are not eligible yet, and enterprises can apply for beta testing.

Section 2: Security and privacy essentials

1. The background proactive research mode is read-only. Modifying content requires verification against custom rules, and sensitive operations such as password changes require manual confirmation.

2. You can view the Activity View work log and intervene to take over. Credentials are used without exposing plaintext passwords.

3. Enterprise workspace content is not used for model training by default; individuals can make their own choice. However, OpenAI reminds that Dots may still make errors, so important work needs to be double-checked.

Section 3: Companion office products

1. ChatGPT Space replaces Library, supporting shared collaboration between teams and AI agents, with built-in Pages comparable to Google Docs and Word.

2. Slides can be generated via natural language, allowing multiple people and AI to edit together; spreadsheets will be added later. Automated workflows can be configured to synchronize project information.

Section 4: Industry updates

1. ChatGPT has 1.2 billion weekly active users. OpenAI plans to raise $30 billion at a $1.4 trillion valuation.

Section 1: New tools for brand marketing

1. Dots can learn a brand's positioning and creative standards. When product information is updated, it can synchronously modify live marketing assets, making it suitable for brands managing multi-channel marketing content.

2. Dedicated Dots can take on email marketing, customer support, and other tasks. Enterprise organizational identity and permissions can be managed, enabling deeper marketing automation.

Section 2: User behavior and trends

1. ChatGPT has 1.2 billion weekly active users. AI agents are becoming a new entry point for user work, so brands should pay attention to how users interact within the AI agent ecosystem.

2. Meta's Muse agent has been downloaded 2.3 million to 4.3 million times in 16 days, with about one-third of users linking business accounts, showing that small and medium merchants and brands are rapidly adopting AI agents.

Section 3: Pricing competition and product strategy

1. OpenAI launched a new $500-per-month ChatGPT Pro tier and adjusted limits on the previous $200 tier, reflecting a tiered subscription strategy for premium users.

2. The release of the lower-cost GPT-6.1 Sol model lowers the barrier to applying AI capabilities, so brands can consider integrating AI capabilities into their product R&D.

Section 1: Market growth and opportunities

1. Always-on AI agents are becoming a new track. Meta's Muse has been downloaded 2.3 million to 4.3 million times in 16 days, topping app store charts and drawing integration applications from over 1,500 companies, signaling explosive demand for commercial applications.

2. Muse has integrated tools such as Shopify, QuickBooks, Stripe, and Canva. About one-third of users linked business accounts, so small and medium sellers can use AI agents to handle store operations and payment scenarios.

Section 2: Applications in sales

1. Dots can match customer needs with product documentation, identify unfinished technical tests, build proof-of-concept solutions, and update proposals as negotiations progress—making it suitable for complex sales processes.

2. OpenAI introduced computer-use capabilities and a plugin ecosystem for the Agents API, allowing developers to customize agents and create new business models.

Section 3: Risks and responses

1. Dots can still make mistakes, so tasks with significant consequences require human review. Sellers using AI agents should set up approval mechanisms and manual confirmation for sensitive operations.

2. The new Pro subscription costs $500 per month; sellers need to evaluate the cost of using AI agents. In addition, Pro users in some European regions do not yet have access, so watch for regional restrictions.

3. OpenAI plans to raise $30 billion at a $1.4 trillion valuation, showing high capital enthusiasm in the AI industry. Sellers can follow the ecosystem of leading platforms to capture opportunities.

Section 1: Product improvement and customer feedback

1. Dots can monitor customer feedback, identify common bugs, run small-scale fix tests, and submit code merge requests with demo videos, helping factory technical teams efficiently collect issues and iterate products.

2. Dots can also be used in R&D scenarios to update analytical conclusions and charts as new experimental data arrives, enabling data analysis for production processes.

Section 2: Digitalization and e-commerce insights

1. Dedicated Dots can take on fixed-role responsibilities such as procurement, invoice processing, customer support, and contract management. Factories can delegate repetitive processes to AI agents to cut labor costs.

2. Automated workflows can sync action items from Slack, email, and calendars, updating owners, deadlines, and next steps in real time—improving coordination efficiency in production projects.

3. The release of the lower-cost GPT-6.1 Sol model lowers the entry threshold for small and mid-sized factories to adopt AI, allowing them to explore integration with business systems.

Section 3: Business opportunities

1. The always-on AI agent market is growing rapidly, with strong demand for third-party tool integrations (more than 1,500 companies applied to integrate with Muse). Factories may consider providing product data access to the AI agent ecosystem or developing supporting tools.

Section 1: New technologies and trends

1. OpenAI released Dots, an always-on AI agent based on GPT-6 Astra. It has an independent cloud computer and browser, and connects to more than 4,000 apps, marking a shift from conversational AI to proactive execution.

2. The Agents API has added computer-use capabilities, event-triggered automation, and deeper Slack and Teams integrations. Service providers can build automation solutions for clients based on these capabilities.

Section 2: Client pain points and solutions

1. Clients worry about AI agent errors and security. OpenAI provides multi-layered permission controls: read-only mode, custom rules, manual confirmation for sensitive operations, Activity View for intervention and takeover, and credential use that does not expose plaintext passwords.

2. Regarding data privacy, enterprise workspace content is not used for training by default; individuals can choose. The agent's background content and private notes are not used for model training.

3. However, OpenAI cautions that Dots may still make mistakes, so service providers should design human review steps into implementations.

Section 3: Competition and business opportunities

1. The always-on AI agent track is highly competitive, and Meta Muse's downloads are growing quickly. Service providers should pay attention to cross-platform agent integration demand.

2. OpenAI has introduced a $500-per-month Pro tier and a low-cost Sol model, creating different service tiers. Service providers can choose according to client budgets.

Section 1: Platform ecosystem openness

1. OpenAI has expanded the Agents API with computer-use capabilities, plugin ecosystem, and event-triggered automation, while deepening Slack and Teams integrations. Platforms can use this as a reference to open their APIs and plugin marketplaces.

2. Dots can connect to over 4,000 third-party apps and interact via web, desktop, mobile, Slack, and Teams; future support for SMS and voice is planned, so platforms need to adapt to multi-channel agent interactions.

Section 2: Business development and partnerships

1. OpenAI has partnered with Microsoft to integrate dedicated Dots into the Microsoft Agent 365 system, allowing enterprises to manage permissions and audits through existing security frameworks. Platforms can learn from this enterprise-grade agent integration model.

2. Meta Muse has integrated with tools such as Shopify, QuickBooks, Stripe, and Canva, and over 1,500 companies have applied for integration. This shows that platform integration with the AI agent ecosystem can bring new users and business scenarios.

Section 3: Operations management and risks

1. Dots provides Activity View work logs, read-only access, and manual confirmation mechanisms. Platform operators should establish similar agent behavior auditing and intervention capabilities.

2. OpenAI has delayed the release of GPT-6.1 Astra due to safety concerns and supports slowing the development of advanced AI models. Platforms should pay attention to AI safety trends and avoid launching high-risk features.

Section 1: New industry dynamics

1. OpenAI's launch of Dots, an always-on AI agent, marks a shift from user-initiated dialogue to continuous background operation and automated software manipulation, driving changes in human-machine collaboration modes.

2. The always-on AI agent space is rapidly crowded: Meta Muse has gained 2.3 million to 4.3 million downloads in 16 days, and other agents such as Instinct, Manus 2.0, Grok Bot, and OpenClaw are emerging, making it a core competitive direction for the AI industry.

Section 2: Emerging issues and policy implications

1. Outside the conference, civil society groups protested, calling for putting public interests first and ending ICE contracts, reflecting controversies over AI ethics and social responsibility.

2. OpenAI disclosed several incidents of unexpected model behavior, delayed the release of GPT-6.1 Astra, and publicly supported slowing down advanced AI development, suggesting regulators need to pay attention to frontier model risks.

3. Dots' multi-layer permission controls and data-training choice mechanisms provide an enterprise-grade product design example for AI agent governance.

Section 3: Business models

1. OpenAI plans a pre-IPO raise of $30 billion at a $1.4 trillion valuation, with annualized revenue of $40 billion, up 70%. It is deferring an IPO to prioritize AI safety, illustrating a capital path for high-risk, high-investment AI companies.

2. Subscription tiering: a $500/month Pro tier, adjusted limits on the $200 tier, and Dots billed separately for additional agents and workload create a monetization model based on usage and feature levels.

Disclaimer: The "Quick Summary" content is entirely generated by AI. Please exercise discretion when interpreting the information. For issues or corrections, please email run@ebrun.com .

I am a Brand Seller Factory Service Provider Marketplace Seller Researcher Read it again.

美国时间2026年9月29日，OpenAI在旧金山Fort Mason举办年度DevDay开发者大会，现场发布超过20项产品及功能更新，核心覆盖常驻AI代理、团队协作办公空间两大产品线。大会举办期间会场外有十余个民间组织联合发起抗议，参与者举着“公众利益优先于利润”“终止ICE合同”等标语，呼吁企业重视AI伦理、暂停向执法及破坏性场景提供技术支持。

本次大会最核心的新品为常驻AI代理Dots，产品基于GPT-6 Astra模型运行，每个代理配备独立云电脑与浏览器，可接入超4000款第三方应用。用户可通过ChatGPT网页端、桌面端、移动端，以及Slack、Microsoft Teams和代理交互，后续还将开放短信、电话语音的沟通渠道。

和传统需要用户主动发起对话的AI助手不同，Dots可在用户关闭聊天窗口后持续运行，同时推进多个项目，根据预设目标监测信息变动、操作软件、完成任务后提交用户审批，过程中会主动同步进度、提出疑问。

现有测试场景覆盖多个岗位工作流程，开发者可让Dots监测客户反馈、定位常见bug、完成小规模修复测试后提交带演示视频的代码合并请求；产品营销团队可让Dots学习品牌定位与创意标准，在产品信息更新后同步修改上线物料；科研场景下Dots可随着新实验数据流入更新分析结论与图表；销售场景下Dots可匹配客户需求与产品文档、识别待完成技术测试项、搭建概念验证方案、随着谈判进展同步更新提案内容；内容团队可让Dots处理访谈转录素材、筛选适合传播的片段、撰写节目笔记与配套社交文案。

目前付费用户可给代理设置名称、自定义卡通形象，产品上线阶段Pro、Business Premium用户可免费获得1个Dots使用权，相关对话不计入日常使用额度，后续将开放多代理配置，额外代理、更高运行速度及更大工作负载将单独计费。欧洲经济区、瑞士、英国的Pro用户暂时无法使用该功能，Enterprise、教育、医疗行业客户可在管理员开通权限后参与beta测试。

Dots设置了多层权限管控机制。后台主动研究模式下代理仅可调用只读工具，无法发送消息、修改内容或操控用户设备。涉及内容修改、外部操作的场景，系统会对照用户设置的自定义规则、平台内置安全要求自动校验，修改密码等敏感操作始终需要人工确认。用户可通过Activity View页面查看所有后台工作记录，随时介入接管操作。Dots调用网站已保存凭证时不会向模型暴露明文密码，用户也可主动授权代理访问个人设备获得更深的操作权限。

针对企业客户，Business、Enterprise、教育版本的工作区内容默认不用于模型训练，个人用户可自主选择是否将Dots相关对话及工作内容用于模型优化，平台不会直接用代理的后台主动研究内容、私有笔记训练模型。OpenAI同时提醒用户，Dots仍有可能出现错误，涉及重要后果的工作需要人工复核。

和Dots同步上线的ChatGPT Space，将替代原有Pro、Business、Enterprise用户的Library功能，成为团队共享协作的承载空间，支持员工、ChatGPT、Codex及Dots在同一上下文下开展工作。Sam Altman在现场演示中提到，Space内的页面和文件和云盘一样集中存储，整个空间可动态运行，用户可给页面设置特定指令，比如巡查团队频道、自动同步最新信息到页面。Space内置Pages文档工具，定位适配人机协作场景，可完成写作、调研、生成图表、创建图片、信息可视化等操作，对标Google Docs、Microsoft Word。

同期上线的协作幻灯片功能支持用户通过自然语言生成演示文稿，多人及AI代理可共同编辑、添加评论，对标Microsoft PowerPoint，后续还将推出协作表格功能。用户可在Space内配置自动化工作流，设定信息抓取来源、内容更新范围及运行频率，让代理持续维护项目页面，从Slack、邮件、日历中同步行动项，实时更新负责人、截止时间及下一步计划，所有内容可随着关联信息变动自动同步。

针对企业客户，OpenAI正在试点专属Dots产品。这类代理拥有独立的组织身份、系统访问凭证及对应工具权限，可承担采购、发票处理、邮件营销、客户支持、商务合同管理等固定岗位职责，目前相关场景已在OpenAI内部完成测试，正联合企业开展定向试点。OpenAI正和微软合作，将专属Dots接入Microsoft Agent 365体系，方便企业通过现有安全管理系统完成权限管控、操作审计。

本次发布的产品线直接切入微软核心的办公软件赛道。作为OpenAI长期深度合作伙伴，过去一年微软、Salesforce等主流软件企业均在密集推出AI相关功能，适配行业变化。Dots所在的常驻AI代理赛道近期已有多个同类产品落地，其中Meta在三周前发布的Muse代理上线后增长迅速，第三方机构统计其上线16天内下载量在230万到430万区间，登顶美国区苹果及谷歌应用商店榜单，目前已新增面向小企业的Shopify、QuickBooks、Stripe、Canva等工具接入，约三分之一用户绑定了商业相关账户，超1500家企业提交了集成申请。同期市场上还有Instinct、Manus 2.0、SpaceXAI的Grok Bot、开源项目OpenClaw等同类产品，常驻AI代理正在成为AI行业新的核心竞争赛道。

本次大会上OpenAI公布，当前ChatGPT周活跃用户规模已达12亿。平台同步推出每月500美元的全新ChatGPT Pro订阅档位，包含最高使用额度及Ultrafast高速响应模式，原每月200美元档位调整使用限额后重新开放。更高阶的GPT-6.1 Astra模型因安全问题暂未发布，此前OpenAI曾披露数起模型非预期行为事件，并在本月公开支持放缓高级AI模型研发节奏、优先管控技术风险的倡议。本次大会同步推出成本更低的GPT-6.1 Sol模型，此外还公布了Agents API的计算机使用能力、扩展插件生态、事件触发自动化、更深层的Slack与Teams集成等更新。

大会举办同日有市场消息称，OpenAI正和投资者洽谈Pre-IPO轮融资，拟募资至少300亿美元，对应估值约1.4万亿美元，该轮融资将作为IPO前的过桥轮。此前OpenAI在2026年3月完成1220亿美元融资，对应估值8520亿美元，原本计划年内上市，目前Sam Altman已明确2026年不会启动IPO，优先推进AI安全相关工作。数据显示，2026年7月以来OpenAI年化营收增长70%，8月年化营收达到400亿美元。

本文首发于亿邦动力官方网站

【版权提示】亿邦动力倡导尊重与保护知识产权。未经许可，任何人不得复制、转载、或以其他方式使用本网站的内容。如发现本站文章存在版权问题，烦请提供版权疑问、身份证明、版权证明、联系方式等发邮件至run@ebrun.com，我们将及时沟通与处理。

已点赞

好文章,值得被更多人看到

FAQ回顾

OpenAI Dots是什么？

Dots是OpenAI于2026年9月发布的常驻AI代理，基于GPT-6 Astra模型运行，每个代理配有独立云电脑和浏览器，可接入超过4000款第三方应用。用户关闭聊天窗口后它仍可持续运行，同时推进多个项目，执行监测信息、操作软件等任务，完成后提交用户审批，并主动同步进度和提出疑问。

常驻AI代理赛道有哪些代表产品？

除OpenAI的Dots外，Meta在三周前发布了Muse代理，第三方机构统计其上线16天内下载量在230万到430万区间，登顶美国区苹果及谷歌应用商店榜单；Instinct、Manus 2.0、SpaceXAI的Grok Bot和开源项目OpenClaw也属于同类产品。常驻AI代理已成为AI行业新的核心竞争赛道。

AI代理Dots可以应用在哪些工作场景？

OpenAI公布的测试场景覆盖多个岗位：开发者可监测客户反馈、定位常见Bug并提交带演示视频的代码合并请求；产品营销团队可学习品牌定位并同步修改上线物料；科研场景可随新数据更新分析结论与图表；销售团队可匹配客户需求、搭建概念验证并更新提案；内容团队可处理访谈素材、撰写节目笔记和社交文案。

ChatGPT Space和传统办公套件有什么区别？

ChatGPT Space是OpenAI推出的团队协作空间，内置Pages文档工具和协作幻灯片功能，支持自然语言生成演示文稿，多人及AI代理可共同编辑；其中Pages对标Google Docs与Word，协作幻灯片对标PowerPoint。与传统办公套件相比，Space还可配置自动化工作流，从Slack、邮件、日历同步信息并实时更新项目状态。

AI代理Dots有哪些安全权限管控措施？

Dots设置多层权限管控：后台主动研究模式只允许调用只读工具；涉及修改和外部操作时会按自定义规则及平台安全要求校验，敏感操作需人工确认；用户可通过Activity View查看记录并接管。调用网站凭证时不暴露明文密码，企业工作区内容默认不用于模型训练，但涉及重要后果的工作仍需人工复核。

标签:: AI 企业办公办公软件\OA 大模型 Chatgpt OpenAI