广告
加载中

风投Khosla押注AI代理Wajo 主打信任安全差异化

亿邦AI 2026-10-08 14:49
亿邦AI 2026/10/08 14:49

邦小白快读

EN
全文速览

总:Wajo是一款主打信任与安全的个人AI代理,帮你代办日常事务。

1. 基础功能:可接入iMessage、WhatsApp,网页卡片展示待办,覆盖日程安排、礼品选购、生活事务,还能代用户打电话给商家或个人,完成预订、取消等操作。

2. 差异化设置:AI完成不了的任务会转给人类处理;生成虚拟信用卡代付,不泄露真实支付信息;用独立邮箱与商家沟通;遇到网站屏蔽AI会即时通知你手动介入或拉群验证。

3. 需要留意:虽然代打电话给航司查询升舱很顺利,但代打给私人联系人时出现过流程不畅、身份披露不清,用起来要有个心理准备。目前产品已在107个国家可用,上线后规模增长10倍,值得尝试。

总:这篇文章给品牌商的重要信号是:AI代理正在成为消费者委托处理日常事务的入口,品牌与用户的交互方式可能被重塑。

1. 用户通过AI代理选礼品、下订单、取消服务,品牌需要让自己的产品和信息能被这类代理高效识别,否则可能被排除在购买流程之外。

2. AI代理代付时使用虚拟信用卡隐藏真实支付信息,品牌将难以掌握消费者真实身份,需思考如何建立信任和售后服务。

3. 用户对大型科技公司用个人数据变现很警惕,品牌若依赖广告平台获取用户,容易遭遇信任危机,更倾向于选择透明、尊重数据的AI代理服务。

4. 早期测试显示AI代理代呼叫人时身份披露不清,品牌客服若接到AI来电,需要明确流程,未来行业或围绕AI代呼叫的身份披露规则产生统一标准。

总:对卖家而言,AI代理是个正在快速放大的新交易入口,但也带来合规和风险挑战。

1. 机会点:Wajo已覆盖107国,上线后规模增长10倍,AI代理正在替用户完成礼品选购、服务预订、取消等动作,卖家应主动研究如何接入这类代理,让商品和服务能被代理推荐和执行。

2. 风险点:亚马逊曾屏蔽同类AI代理Muse,Wajo宣称架构符合亚马逊服务条款,说明平台对AI代理访问有严格限制,卖家在对接代理时需要确保操作符合平台规则,否则可能被下架或处罚。

3. 信任属性影响消费:用户选择AI代理时优先看中信任和安全,卖家在详情页、客服话术上要透明,不能诱导AI代理做出错误决策。

4. 学习点:Wajo的做法值得借鉴,包括虚拟卡代付保护隐私、AI搞不定时转人工,这些都能增强消费者安全感,卖家服务也可引入类似的人机协作机制。

总:这篇文章对工厂的最大启示是:AI代理把信任与安全做成产品架构,而不是后期补丁,这为产品设计和数字化转型提供了新方向。

1. 产品设计中,安全与隐私必须从架构第一天就考虑,Wajo用虚拟卡隐藏信息、可信联系人限定呼叫范围,这些设计理念可直接迁移到智能硬件或配套服务中。

2. 商业机会:AI代理能代打电话、代发邮件、代付,工厂如果做直接面向消费者的产品,可考虑预留API或交互接口,让AI代理能帮用户完成订单查询、售后报修等操作,提升服务效率。

3. 数字化启发:Wajo在AI无法完成任务时引入人类接手,说明自动化不是100%,工厂的数字化系统也应设计人工兜底环节。

4. 数据管理:文章提到模型遗忘技术未攻克,用户数据一上传就永久留存,工厂在打造电商或数字平台时,必须设计可删除、可撤回的数据流程,否则会成为信任短板。

总:AI代理赛道竞争激烈,服务商的机会在于解决信任安全、人机协作、身份披露等痛点。

1. 行业趋势:个人AI代理每周都有新竞品入场,信任属性成为风投下注关键,Wajo把信任安全放在第一位。

2. 新技术方向:虚拟信用卡系统隐藏用户真实支付信息;独立邮箱进行邮件沟通;可信联系人列表限制呼叫对象;用户可永久记录拒绝来电偏好,这些都是可落地的新技术方案。

3. 客户痛点:AI代理在代打私人电话时身份披露不清晰、流程不畅;模型遗忘未解决导致数据永久留存;网站或服务会屏蔽AI代理访问,这些是服务商可以切入的优化点。

4. 解决方案:辅以人类客服兜底,AI搞不定就直接转人工;遇到屏蔽时通知用户手动介入或拉起三方群组验证;围绕AI代呼叫制定身份披露规则,未来可能要成为行业标配。

总:平台商需要高度关注AI代理的兼容性与规则风险,避免被代理绕过或制造合规漏洞。

1. 平台现状与问题:AI代理会代替用户访问网站、发邮件、打电话、支付,平台无法判断背后是真人还是代理,可能出现服务滥用或体验失控;亚马逊已屏蔽过同类产品Muse,Wajo自称完全符合亚马逊服务条款,说明平台正在用服务条款来约束代理行为。

2. 按平台做法:平台可要求AI代理遵守服务条款,对不符合的访问进行屏蔽;也可以学习Wajo的做法,在遇到屏蔽时通知用户介入或拉起包含用户、商家的群组验证,使身份透明化。

3. 招商与合作:AI代理生态刚刚爆发,平台可以主动推出针对代理访问的开放接口或白名单机制,既能吸引优质AI代理服务,又能控制风险。

4. 运营与风控:用户数据信任是敏感点,平台不能沿用广告变现逻辑,需要建立符合隐私保护的数据处理流程;可信联系人列表和永久拒绝来电偏好等机制,可为平台的风控规则提供参考。

总:这篇文章呈现了个人AI代理产业的一个关键转折:信任安全被提升到核心竞争力,并由此引发一系列新问题。

1. 产业新动向:AI代理竞争从拼功能转向拼信任,风投明确把信任属性作为加仓依据;产品在107国增长10倍,说明个人代理有真实需求。

2. 新问题:模型遗忘技术尚未攻克,用户数据一旦上传可能永久留存;AI代呼叫私人联系人时身份披露规则不清晰;平台屏蔽AI代理的合规边界模糊。

3. 政策法规启示:应建立AI呼叫身份强制披露规则,允许用户设置可信联系人范围并永久记录拒绝来电请求;对虚拟信用卡等隐藏支付信息的技术,需要监管与法律协调;数据保留应有期限,不能靠用户第一天的选择来决定永久留存。

4. 商业模式:AI代理与人类混合服务(AI无法完成转人工)可能成为重要模式,虚拟卡代付、独立邮箱等设计为代理商业化提供了隐私保护底座,值得深入研究其成本与可持续性。

返回默认

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

Overall: Wajo is a personal AI agent focused on trust and security, helping users handle everyday tasks.

1. Core features: It can connect to iMessage and WhatsApp, display to-dos in web cards, cover scheduling, gift selection, and life admin, and make calls to businesses or individuals on the user's behalf to complete bookings, cancellations, and similar tasks.

2. Differentiation: Tasks the AI cannot complete are escalated to humans; virtual credit cards are generated for payments without exposing real payment information; dedicated email addresses are used for merchant communication; and if a website blocks the AI, the user is notified in real time to intervene manually or join a group verification.

3. Caveats: The test call to an airline about upgrades went smoothly, but calls to private contacts had workflow issues and unclear identity disclosure, so users should set expectations. The product is now available in 107 countries and has grown 10x since launch—worth trying.

Overall: For brands, the key signal is that AI agents are becoming the gateway through which consumers delegate daily tasks, and the way brands interact with users may be reshaped.

1. Users rely on AI agents to select gifts, place orders, and cancel services. Brands need to make their products and information easily identifiable to such agents, or risk being excluded from the purchase journey.

2. Because AI agents use virtual cards to pay while hiding real payment information, brands will find it harder to identify the actual consumer. They need to think about how to build trust and handle after-sales support.

3. Users are wary of big tech companies monetizing personal data. Brands that depend on ad platforms to acquire users may face a trust crisis and are more likely to favor transparent, data-respecting AI agent services.

4. Early tests show that AI-agent calls have unclear identity disclosure. If brand customer service receives AI-initiated calls, clear procedures are needed. The industry may eventually develop uniform standards for identity disclosure in AI-placed calls.

Overall: For sellers, AI agents are a rapidly expanding new transaction gateway, but they also bring compliance and risk challenges.

1. Opportunity: Wajo covers 107 countries and has grown 10x since launch. AI agents are already helping users with gift selection, service bookings, and cancellations. Sellers should actively study how to integrate with such agents so their products and services can be recommended and executed.

2. Risk: Amazon previously blocked a similar AI agent, Muse. Wajo claims its architecture complies with Amazon's terms of service, indicating that platforms impose strict limits on AI agent access. Sellers must ensure their integrations follow platform rules, or risk delisting or penalties.

3. Trust attributes affect purchasing: Users prioritize trust and security when choosing an AI agent. Sellers should be transparent in product pages and customer-service scripts and avoid inducing AI agents into incorrect decisions.

4. Lessons: Wajo's practices are worth learning from, including virtual cards for payment privacy and human escalation when the AI cannot complete a task. Sellers can introduce similar human-machine collaboration in their own services.

Overall: The key lesson for factories is that Wajo builds trust and security into the product architecture from day one rather than treating them as later patches—this points to a new direction for product design and digital transformation.

1. In product design, security and privacy must be considered from the first day. Wajo hides user information with virtual cards and restricts call targets through trusted contacts; these design ideas can be transferred to smart hardware or supporting services.

2. Business opportunity: AI agents can make phone calls, send emails, and make payments. If a factory sells directly to consumers, it can reserve APIs or interaction interfaces so AI agents can help users with order inquiries, after-sales repairs, and other operations, improving service efficiency.

3. Digital transformation insight: Wajo introduces humans when the AI cannot complete a task, showing that automation is not 100%. Factory digital systems should also build in human fallback mechanisms.

4. Data management: The article notes that machine unlearning has not been solved, so uploaded user data may remain permanently. Factories building e-commerce or digital platforms must design data flows that allow deletion or withdrawal; otherwise, trust will become a weak point.

Overall: The AI agent space is highly competitive. The opportunity for service providers lies in solving pain points around trust and security, human-machine collaboration, and identity disclosure.

1. Industry trend: New competitors in personal AI agents are entering weekly, and trust attributes have become a key factor in venture investment. Wajo puts trust and security first.

2. New technical directions: virtual credit card systems hide real payment information; dedicated email addresses handle communication; trusted contact lists restrict who the agent can call; and users can permanently record call rejection preferences. These are practical solutions that can be implemented now.

3. Customer pain points: AI agents have unclear identity disclosure and poor workflow when calling private contacts; machine unlearning remains unresolved, so data persists indefinitely; and websites or services may block AI agent access. These are optimization points for service providers.

4. Solutions: use human-agent fallback so unresolved tasks are routed to people; notify users to intervene manually or set up three-party group verification when blocked; and develop identity disclosure rules for AI-initiated calls, which may become an industry standard.

Overall: Platform operators need to pay close attention to AI agent compatibility and rule risks, avoiding bypasses or compliance loopholes.

1. Current situation and problems: AI agents access websites, send emails, make calls, and process payments on behalf of users. Platforms cannot tell whether the actor is a real person or an agent, creating risks of service abuse or experience loss of control. Amazon previously blocked the similar product Muse; Wajo claims it fully complies with Amazon's terms of service, showing that platforms are using terms to constrain agent behavior.

2. Platform countermeasures: Platforms can require AI agents to comply with their terms of service and block noncompliant access. They can also learn from Wajo by notifying users when blocked, or creating group verification involving the user and merchant, making identity more transparent.

3. Ecosystem development and cooperation: The AI agent ecosystem is just emerging. Platforms can proactively offer open APIs or whitelist mechanisms for agent access, attracting high-quality AI agents while controlling risk.

4. Operations and risk control: User data trust is a sensitive issue. Platforms cannot simply follow advertising-monetization logic; they need to establish privacy-respecting data processing flows. Mechanisms like trusted contact lists and permanent do-not-call preferences can serve as references for platform risk-control rules.

Overall: This article shows a key turning point in the personal AI agent industry: trust and security have become core competitive factors, triggering a series of new questions.

1. New industry dynamics: Competition among AI agents is shifting from features to trust. VCs explicitly cite trust attributes as a reason to increase investment. The 10x growth in 107 countries shows real demand for personal agents.

2. New questions: machine unlearning has not been solved, so uploaded data may persist forever; identity-disclosure rules for AI calls to private contacts are unclear; and the compliance boundaries for platforms blocking AI agents remain ambiguous.

3. Policy and regulatory implications: mandatory identity-disclosure rules for AI-initiated calls should be established; users should be allowed to set trusted-contact ranges and permanently record call rejection requests; virtual-card and other payment-hiding technologies need regulatory and legal coordination; and data retention should have time limits rather than being permanently decided by users' first-day choices.

4. Business models: hybrid AI-human services, where humans take over when AI cannot complete tasks, are likely to become an important model. Virtual-card payments and dedicated email addresses provide a privacy-preserving base for agent commercialization, and their costs and sustainability deserve deeper research.

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.

2026年10月,个人AI代理赛道竞争持续升温,Meta旗下Muse、OpenAI旗下Dot、Instinct等产品占据先发位置,每周仍有新的竞品进入市场。硅谷顶级风投人Vinod Khosla公开站队前谷歌、DeepMind工程师Shivani Poddar创办的AI代理产品Wajo,核心判断依据为产品的信任属性。

AI代理的核心价值是替用户执行各类操作,信任是产品体验的基础。Khosla在对外通话中对比Muse、Instinct等竞品,Wajo从架构设计之初就将信任与安全放在首位,同类产品大多未做到这一点。Meta核心营收来自广告业务,用户上传个人数据后会成为平台变现的载体而非服务对象,多数用户对这类大型科技公司存在数据信任顾虑。

Wajo核心产品Fo的基础功能与市面主流AI代理大致对齐。产品可接入iMessage、WhatsApp渠道,网页端以卡片形式分类展示待办任务,覆盖日程安排、礼品选购、生活事务处理等类目,支持代用户致电商家或个人,完成服务预订、取消等操作。早期测试过程中,产品可顺利完成致电航司查询升舱权益等任务,在代用户致电私人联系人提醒事项的场景中,曾出现流程不畅、身份披露不清晰的问题。目前Instinct、Muse等竞品也在逐步开放代呼叫能力,后续行业或将围绕AI代呼叫场景的身份披露规则形成更多讨论。Fo有一项差异化设计,遇到AI无法独立完成的任务时,可直接聘用人类接手完成。

Fo还支持生成虚拟信用卡完成代付,全程不泄露用户真实支付信息。产品配备独立邮箱,可直接与个人、商家完成邮件沟通。遇到网站或服务屏蔽AI代理访问的情况,会第一时间通知用户手动介入,或拉起包含用户、商家的沟通群组完成身份验证。

Wajo已拿到谷歌DeepMind负责人Jeff Dean、硅谷知名天使投资人Gokul Rajaram的投资,暂未披露具体融资金额。官方未公开累计用户规模,仅透露产品目前已在107个国家投入使用,自9月末正式上线以来规模增长10倍。

创办Wajo之前,Poddar曾在Meta、谷歌、DeepMind历任多个工程岗位,参与打造初代Facebook助手。参与Gemini落地及配套安全体系搭建的经历,让她意识到AI代理产品需要足够的技术深度支撑,谷歌这类技术公司往往受机制影响出现产品落地滞后的问题,她希望打造同时兼顾技术能力与产品体验、贴合用户真实需求的产品。

她将信任作为Wajo的核心设计原则。目前AI代理领域尚未攻克“模型遗忘”技术难题,产品如果在上线首日就获取用户个人数据,这类数据将被永久留存。信任构建必须从产品上线第一天启动,覆盖技术架构到产品选择的全流程。

目前Wajo已围绕信任与安全落地多项具体设计,正在为可隐藏用户信息的虚拟卡系统申请专利。针对呼叫场景,产品将上线“可信联系人”列表,限定AI代理可呼叫的对象范围,接到AI呼叫的用户可直接提出不再接收相关来电的要求,系统将永久记录这类偏好。针对此前Muse被亚马逊屏蔽的问题,Wajo称其产品架构完全符合亚马逊的服务条款,不会触发同类合规问题。按照产品规划,未来几个月团队将探索Fo代理在多用户社交场景下的应用模式,全程保留隐私保护的核心设计。

本文首发于 亿邦动力 官方网站

文章来源:亿邦动力

广告
微信
朋友圈

FAQ回顾

个人AI代理是什么?为什么信任和安全很重要?

个人AI代理是替用户执行日程安排、礼品选购、服务预订、代打电话等操作的应用,信任是产品体验的基础。Wajo创始人Shivani Poddar认为,AI代理需从第一天构建信任,因为模型遗忘难题尚未攻克,数据一旦留存可能永久保留;同时要避免用户数据被平台变现或滥用,所以信任与安全必须从技术架构和产品选择的全流程落实。

Wajo是什么?它有哪些独特功能?

Wajo是前谷歌、DeepMind工程师Shivani Poddar创办的个人AI代理产品,核心产品为Fo。它可接入iMessage、WhatsApp,通过网页端卡片管理日程、礼品、生活事务,支持代用户致电商家完成预订或取消;差异化功能包括AI无法独立完成时聘用人类接手、生成虚拟信用卡代付、配备独立邮箱,以及遇到网站屏蔽时通知用户或拉起群组验证。

AI代理的信任问题具体指什么?Wajo如何解决?

AI代理需要获取个人数据执行操作,但目前存在模型遗忘难题,上线首日获取的数据可能被永久留存。Wajo从架构设计之初将信任与安全放在首位,已落地多项设计:为可隐藏用户信息的虚拟卡系统申请专利,即将上线限定AI呼叫范围的“可信联系人”列表,用户可要求永久不再接收来电;同时称产品架构完全符合亚马逊服务条款,不会触发类似Muse被屏蔽的合规问题。

个人AI代理赛道有哪些代表产品和趋势?

截至2026年10月,个人AI代理赛道代表产品包括Meta旗下Muse、OpenAI旗下Dot、Instinct,以及主打信任安全的Wajo。趋势方面,Instinct、Muse等竞品正逐步开放代呼叫能力,行业将围绕AI代呼叫场景的身份披露规则形成更多讨论;同时,Wajo等新创产品以隐私信任作为差异化竞争点,已覆盖107个国家,自9月末上线以来规模增长10倍。

Khosla为什么投资Wajo?他看好什么?

硅谷知名风投Vinod Khosla公开支持Wajo,核心判断依据是产品的信任属性。他在对外通话中对比Muse、Instinct等竞品,认为多数同类产品未将信任与安全放在架构设计的首位;例如Meta核心营收来自广告,用户上传个人数据后可能成为平台变现载体,而Wajo从第一天起将用户服务而非变现作为目标,这种信任导向正是他看重的差异化价值。

这么好看,分享一下?

朋友圈 分享

APP内打开

赞 +1
+1
微信好友 朋友圈 新浪微博 QQ空间
关闭
收藏成功
发送
/140 0