章节 1：Sierra 的使命——跨越企业客服体验的鸿沟

📝 本节摘要：

本节中，Clay Bavor 首先介绍了 Sierra 的核心愿景：利用 AI 帮助企业打破“高品质服务”与“高昂成本”之间的不可能三角。他通过生动的日常案例（路人对着电话大喊“转人工”）指出了传统客服的痛点。随后，他列举了 ADT 和 SiriusXM 等客户案例，展示了 AI 智能体如何实际解决复杂问题（如发送卫星信号），并提出核心观点：“品牌智能体”将是继网站和 App 之后的下一代企业交互形态。

[原文] [Alessio Fanelli]: so welcome uh I I know a lot of people might be familiar familiar with you and especially all the amazing work you've done at Google in the past and with Sierra Sierra overall but maybe just give a one minute intro on what Sierra is who you serve the scale of it

[译文] [Alessio Fanelli]: 欢迎，我知道许多人可能对你很熟悉，特别是你在谷歌过去所做的出色工作以及 Sierra 的整体情况，但也许可以花一分钟介绍一下 Sierra 是什么，你们服务谁，以及它的规模？

[原文] [Clay Bavor]: yeah uh so Sierra in a nutshell we help businesses build better more human customer experiences with AI and uh concretely what we're trying to do is bridge this age-old gap between businesses wanting to provide great care customer service customer experience on the one hand and the impossibility of doing that on the other hand because of cost

[译文] [Clay Bavor]: 好的，简而言之，Sierra 利用 AI 帮助企业构建更好、更人性化的客户体验。具体来说，我们试图弥合一个古老的鸿沟：一方面是企业想要提供优质的关怀、客户服务和体验，另一方面是由于成本原因导致这在实际上是不可能的。

[原文] [Clay Bavor]: and I think we've all been there just like being on hold was like 000000 or uh I was walking into work a couple weeks ago and there was a dude uh on a call holding his iPhone with AirPods in yelling representative representative representative representative and um like I'm not sure what circle of hell involves you know waiting on hold and and trying to get a problem solved but I'm sure it's one

[译文] [Clay Bavor]: 我想我们都有过这样的经历，就像电话等待时按 000000，或者几周前我走在上班路上，看到一个家伙在打电话，拿着 iPhone 戴着 AirPods 大喊“人工代表、人工代表、人工代表、人工代表”。我不确定等待接听并试图解决问题属于地狱的哪一层，但我肯定它是其中之一。

[原文] [Clay Bavor]: and we're trying to uh bridge that gap and so uh concretely you know we're a year in change after launch we have hundreds of customers we'll serve hundreds of millions of consumers this year

[译文] [Clay Bavor]: 我们正试图弥合这一鸿沟。具体来说，我们上线一年多了，拥有数百家客户，今年我们将服务数亿消费者。

[原文] [Clay Bavor]: Um uh we work with folks like ADT the largest home security company in the country built an AI for them if any of you have a SiriusXM subscription and contact them you'll speak with Harmony an AI agent that we built with and for them who picks up the phone and can help you uh get back up and running including sending a satellite signal from space um to to get your radio up and running again

[译文] [Clay Bavor]: 我们与像 ADT 这样的伙伴合作，它是国内最大的家庭安全公司，我们为他们构建了一个 AI；如果你订阅了 SiriusXM 并联系他们，你会与 Harmony 交谈，这是我们与他们共同构建并为他们打造的 AI 智能体，它会接听电话并帮助你恢复服务，包括从太空发送卫星信号让你的收音机重新运作。

[原文] [Clay Bavor]: we work with uh one of the largest mortgage originators in the country and then a lot of uh local tech companies that you would have heard about and so um that's that's what we do in in a nutshell

[译文] [Clay Bavor]: 我们与国内最大的抵押贷款发起机构之一合作，还有许多你可能听说过的本地科技公司，简而言之这就是我们所做的。

[原文] [Clay Bavor]: and um uh what excites us is we think that every company in the future is going to have an agent its own uh branded customerf facing agent and we think it's what comes after the website what comes after the mobile app and we want to help the the great companies of the world build their own and do it well

[译文] [Clay Bavor]: 令我们兴奋的是，我们认为未来的每家公司都将拥有一个智能体，即其自有的、面向客户的品牌智能体。我们认为这是继网站、继移动应用程序之后的事物，我们希望帮助世界上伟大的公司构建它们自己的智能体并将其做好。

章节 2：AI 架构师崛起——AI 时代的“Webmaster”

📝 本节摘要：

本节探讨了一个随 AI 普及而诞生的新兴关键角色——“AI 架构师”。Clay Bavor 将其类比为互联网早期的“网站管理员（Webmaster）”，即负责企业数字门面的核心人物。他详细拆解了该角色的三大核心能力要素：对技术能力的感知、对品牌调性与体验的审美把控（如 Chubbies 的搞怪智能体 Duncan Smothers），以及对商业成果的驱动力。令人意外的是，这些人才往往并非纯技术背景，而是更多来自客服与客户体验（CX）团队。

[原文] [Alessio Fanelli]: so talking about AI architect I would say the killer use cases of AI are coding and customer support today um the coding I would guess vibe coding and some of these ideas are kind of t taking hold on the customer support side you know Brett mentioned this idea of the AI architect where instead of managing software you're almost like the personality coach and like really thinking through what's the vibe that the agent should have how does it integrate so what is an AI architect

[译文] [Alessio Fanelli]: 说到 AI 架构师，我想说目前 AI 的杀手级应用是编程和客户支持。在编程方面，我想像“氛围编码（vibe coding）”这类概念正在流行；而在客户支持方面，Brett 提到了 AI 架构师这个概念，在这里你不再是管理软件，而是几乎像一个性格教练，真正去思考智能体应该具备什么样的氛围，它如何整合。那么，什么是 AI 架构师？

[原文] [Clay Bavor]: so around the emergence of the internet and the web there was this role of web master right and you don't really hear web master thrown around a lot these days but it was someone who was creating a company's in essence digital storefront right if if they're a business and thinking about not only the technology right uh uh are you building an ASP or you have static pages or or whatever in the dark ages of the web uh but also it's like what does it look like what does it feel like and and so on

[译文] [Clay Bavor]: 在互联网和万维网出现之初，有一个角色叫“网站管理员（Webmaster）”，对吧？如今你不太听到这个词了，但在当时，如果是一家企业，实际上就是这个人在打造公司的数字门面。他们不仅要考虑技术——比如在互联网的黑暗时代，你是构建 ASP 还是静态页面等等——还要考虑它看起来怎么样、感觉怎么样等等。

[原文] [Clay Bavor]: so I think the AI architect I would say is kind of the AI era and AI agent version of the web master in a way and it's um it's actually Brett shared on the podcast it's a role that uh we've seen emerge organically across a set of our customers and we heard it first from one of our largest customers where the the team of folks who are responsible for managing coaching improving building their agent came to calling themselves the AI architect

[译文] [Clay Bavor]: 所以我认为，AI 架构师在某种程度上就是 AI 时代和 AI 智能体版本的“网站管理员”。这实际上正如 Brett 在播客中分享的，这是我们在许多客户中看到自然涌现出的一个角色。我们最早是从我们最大的一个客户那里听到的，那里负责管理、指导、改进和构建智能体的团队开始自称为 AI 架构师。

[原文] [Clay Bavor]: and I think there are three parts to it one is you got to understand the the technology have a little bit of a feel for what agents can do and so doesn't mean that you've you know pre-trained a a trillion parameter model or uh even been hands-on with a lang chain or um even vibe coding but having a feel for the capabilities number one

[译文] [Clay Bavor]: 我认为这包含三个部分。第一，你必须理解技术，对智能体能做什么有一点感觉。这并不意味着你预训练过万亿参数的模型，也不意味着你亲自操作过 LangChain 甚至进行过“氛围编码”，而是要对（模型）能力有感觉，这是第一点。

[原文] [Clay Bavor]: number two is and a pattern we've seen we some interesting things is a company's agent needs to not only manifest the functional capabilities of the company but also be something of a brand ambassador right what's the voice what are the values what's the tone how do you come across how do you create a connection with the customer so there's a real I would say aesthetic and taste element to it how should it sound uh does it have a persona right

[译文] [Clay Bavor]: 第二点，也是我们看到的一个有趣模式，就是一家公司的智能体不仅需要体现公司的功能性能力，还需要成为某种品牌大使，对吧？它的声音是什么？价值观是什么？语调是什么？你如何表现？如何与客户建立联系？所以我认为这里面确实有一个审美和品味的要素。它听起来应该是什么样的？它有人设吗？

[原文] [Clay Bavor]: some of our customers agents are the ex company virtual agent Uh in contrast we work with a company called Chubbies uh they make great very short shorts uh that I am not cool enough to wear um their agent is named Duncan Smothers and will tell kind of irreverent broy jokes and so on and speak in a a really funny tone so making decisions about that and and many other ways that the agent comes across is the second part of it

[译文] [Clay Bavor]: 我们有些客户的智能体就是“某某公司虚拟代理”。相比之下，我们与一家叫 Chubbies 的公司合作——他们生产很棒的超短裤，我还没酷到敢穿那种——他们的智能体叫 Duncan Smothers，会讲那种玩世不恭的兄弟式笑话，用一种非常有趣的语调说话。所以，决定这些以及智能体表现形式的许多其他方面，是第二部分。

[原文] [Clay Bavor]: and then third ultimately a business wants to in engaging with its customers drive business outcomes so what business outcomes are you driving towards so it's this three hats it's technology it's experience and aesthetics and design and it's business and I I think it will be one of the fastest growing job types in the next 5 years

[译文] [Clay Bavor]: 第三点，归根结底，企业希望通过与客户互动来推动业务成果。那么你的目标是什么业务成果？所以这是三顶帽子：技术、体验与审美设计、以及业务。我认为这将是未来 5 年增长最快的工作类型之一。

[原文] [Clay Bavor]: and from what I've seen you know a front row seat on this one of the most interesting as well where most of them already in the customer support org or are some of these people coming from more technical teams and kind of like creating this blend of a role all of the above but the the area I've been most excited about is seeing individuals in customer experience teams

[译文] [Clay Bavor]: 根据我所看到的——我就坐在前排观察这一切——这也是最有趣的一点：这些人大部分已经在客户支持部门了，还是来自更偏技术的团队并创造了这种混合角色？两者都有，但我最兴奋的是看到客户体验团队中的个人。

[原文] [Clay Bavor]: and you know engineering teams are often celebrated and held up and so on your your CX team less often so uh but what's emerged are people who really do have a feel for what a great customer experience looks like and are hands-on with the technology enough and have a sense for what the business is trying to do so the answer is folks emerge from all of those teams i would say the most common and the one I'm most excited about are the folks who have been close to the customer in service support care retailing settings and so on who uh kind of put this badge on and and become the the AI architect or one of the AI architects for their for their business

[译文] [Clay Bavor]: 你们知道，工程团队经常被赞颂和推崇，而客户体验（CX）团队则较少如此。但是现在涌现出的一批人，他们真正对什么是卓越的客户体验有感觉，并且能够足够熟练地操作技术，同时也了解企业想要做什么。所以答案是，人才从所有这些团队中涌现，但我得说最常见、也最让我兴奋的，是那些在服务、支持、关怀、零售等场景中贴近客户的人，他们戴上了这枚徽章，成为了他们企业的 AI 架构师。

章节 3：落地策略——拥抱风险与重构服务团队

📝 本节摘要：

在本节中，Clay Bavor 分享了成功落地 AI 策略的企业所具备的关键特质。首先，他强调“不要让完美成为优秀的敌人”，因为大语言模型本质上是概率性软件，企业必须具备一定的风险承受能力和探索精神。其次，他建议从具体的业务痛点入手（例如处理单一退货流程），而不是空谈宏大的 AI 战略。最后，他提出了“团队重构”的概念：成功的企业不会将旧的团队结构生搬硬套到 AI 时代，而是会建立全新的角色（如专门审查对话并“指导”智能体的团队），从第一性原理出发重新设计客户体验组织。

[原文] [Alessio Fanelli]: i'm sure there's a lot of people in the audience that have been tasked to figure out the AI strategy of their company whatever whatever that means the board says we need an really angry um so when you think about all the AI architects that you work with what are like some of the traits of the most successful one are they really curious do they try a lot of products are they very structured in how they evaluate are they maybe more bite- based on how they think about what tools to use i mean there's a lot in what what has made customers we've worked with successful in developing an AI strategy and actually applying it

[译文] [Alessio Fanelli]: 我敢肯定观众席里有很多人被指派去弄清楚公司的 AI 战略，不管那到底意味着什么——董事会说我们需要一个非常愤怒的（注：此处原文疑为口误或中断，直译保留）……嗯，所以当你想到所有和你合作过的 AI 架构师时，那些最成功的人有哪些特质？他们是否非常好奇？他们是否尝试很多产品？他们在评估时是否非常有条理？或者他们在考虑使用什么工具时是否更基于碎片化信息？我的意思是，对于那些在制定 AI 战略并实际应用方面取得成功的客户来说，其中有很多因素。

[原文] [Clay Bavor]: i I think um broadly across the businesses we work with the most successful have not let perfect be the enemy of the good you think about large language models and agents these are probabilistic pieces of software that could say or do anything and so necessary in adopting them is some amount of risk tolerance and being willing to you know step into the pond and and try things out so a spirit of exploration trying new things and taking some risk uh is is number one

[译文] [Clay Bavor]: 我认为，从我们合作的广泛企业来看，最成功的那些没有让完美成为优秀的敌人。你想想大语言模型和智能体，它们是概率性的软件，可能会说任何话或做任何事。因此，采用它们必须具备一定程度的风险承受力，并且愿意——你知道的——踏入池塘去尝试事物。所以，一种探索精神、尝试新事物并承担一些风险，这是第一位的。

[原文] [Clay Bavor]: number two is a deep uh focus on actually solving customer problems and real business problems i think uh too often there's this hey let's apply some AI to that and we'll have you know emerge from that our AI strategy no no no no like start with a concrete valuable problem to solve and it can be very narrow uh you know we to give you a sense for like where we and one of our customers started was something as simple as processing a single return

[译文] [Clay Bavor]: 第二点是深度聚焦于实际解决客户问题和真正的业务问题。我认为太常见的情况是“嘿，让我们在那上面应用一些 AI，然后我们的 AI 战略就会从中浮现出来”。不不不，要从解决一个具体、有价值的问题开始，它可以非常狭窄。给你一个概念，比如我们和一个客户起步的地方，就是简单到处理单次退货这样的事情。

[原文] [Clay Bavor]: and like we celebrated and they celebrated when their AI picked up the phone and successfully drop shipped someone a new pair of shoes and printed and gave them a shipping label to print right it is not you know the pinnacle of complexity and so on but you start somewhere um and and learn and grow from there

[译文] [Clay Bavor]: 当他们的 AI 接听电话并成功给某人代发了一双新鞋，并提供给他们一张可以打印的运输标签时，我们庆祝了，他们也庆祝了。对吧？这并不是复杂性的巅峰之类的，但你必须从某个地方开始，然后从那里学习和成长。

[原文] [Clay Bavor]: and then the last thing I would say is uh not shoehorning the way you've built teams in the past or done things in the past into the AI era and so our most successful uh customers and partners have actually rearchitected their customer experience and customer service teams around supporting the AI agent in being and doing better

[译文] [Clay Bavor]: 最后我要说的是，不要把你过去建立团队的方式或做事的方式生搬硬套到 AI 时代。我们最成功的客户和合作伙伴实际上已经重构了他们的客户体验和客户服务团队，围绕着支持 AI 智能体变得更好、做得更好来进行。

[原文] [Clay Bavor]: so there's a set of people for instance at one of our customers who will review a couple hundred conversations a day and basically coach and refine the agent on how to do it better how to say it better how to make better decisions how to have greater empathy how to have better judgment and that was not a team that has existed you know anywhere into the past and so really thinking from first principles and and not just trying to translate naively the old to the new would be a third element of it

[译文] [Clay Bavor]: 比如在我们的一位客户那里，有一群人每天会审查几百次对话，基本上是指导和完善智能体如何做得更好、如何表达得更好、如何做出更好的决定、如何拥有更强的同理心、如何拥有更好的判断力。这不是一个过去存在于任何地方的团队。所以，真正从第一性原理出发思考，而不是仅仅试图天真地将旧事物转化为新事物，这将是第三个要素。

章节 4：自建 vs 购买——揭秘“智能体冰山”理论

📝 本节摘要：

本节中，Clay Bavor 深入探讨了企业在 AI 领域面临的经典“自建还是购买”难题。他提出了“智能体冰山”理论，指出工程团队往往只关注到冰山一角（如模型选择、向量数据库），却忽视了水面下巨大的隐形复杂度（如回归测试、模型迁移、语音打断处理等）。他介绍了 Sierra 的解决方案 Agent OS，这是一个结合了深层代码构建与无代码工具的平台，不仅解决了技术难题，还赋予了非技术人员直接优化智能体的能力。Clay 指出，许多试图自建的企业最终因低估复杂度而不得不重新寻求合作。

[原文] [Alessio Fanelli]: you know architect that's kind of like a enter technical connotation in enterprises like hey you're like the software architect what were some of the build versus buy fallacies that you've seen working with customers where you maybe have the customer support team that just wants something today and the engineering team is like "Oh we can just build this it's just going to take three times as long and cost twice as much as buying." The multiples are more than that but yes

[译文] [Alessio Fanelli]: 你知道“架构师”在企业中带有一种技术内涵，就像“嘿，你是软件架构师”。你在与客户合作时见过哪些关于“自建还是购买”的谬误？也许客户支持团队今天就想要某个东西，而工程团队却说：“哦，我们可以自己建这个，只不过要花三倍的时间和两倍的成本。”虽然实际倍数远不止于此，但这确实是个问题。

[原文] [Clay Bavor]: yeah so it's such an important question and uh we get the oh we're going to build our own uh you know why should we work with you all the time and it's funny we we have a slide we call it the agent iceberg where I think technical teams think oh awesome we're going to choose our language model should we Langraph or Langchain take it off the shelf uh you know what uh what embeddings model will we use uh what vector database and you know maybe we'll integrate some tools you know we're we're done

[译文] [Clay Bavor]: 是的，这是个非常重要的问题。我们经常听到“哦，我们要自己建，为什么要和你们合作”这类话。有趣的是，我们有一张幻灯片叫“智能体冰山”。我认为技术团队会觉得：“哦太棒了，我们要选语言模型了，是用 LangGraph 还是 LangChain？直接拿现成的。我们用什么嵌入模型？什么向量数据库？也许再集成一些工具，我们就搞定了。”

[原文] [Clay Bavor]: and then you put on your scuba tank and you kind of you go under the surface of the ocean like oh my god you know there there are hundreds of things from how do you do regression testing and unit testing how do you do model migration and model upgrades

[译文] [Clay Bavor]: 然后当你背上水肺氧气瓶潜入海面之下，你会惊呼：“天哪！”你会发现下面有成百上千件事：比如如何做回归测试和单元测试？如何做模型迁移和模型升级？

[原文] [Clay Bavor]: uh in voice it gets just shockingly complicated where how do you separate primary from secondary speaker handle interruptions and a thousand other things in the agent development life cycle

[译文] [Clay Bavor]: 在语音方面，情况会变得极其复杂：你如何区分主讲人和次讲人？如何处理打断？以及智能体开发生命周期中的其他一千件事。

[原文] [Clay Bavor]: and so um we we come to our customers with uh what we call agent OS our platform for building agents and it's it's a very sophisticated toolkit for incode building very sophisticated customer-f facing agents now the the architect right there's this whole other side to building excellent customerf facing agents which is the experiential the the brand the market so paired with that we have a set of noode tools that enable nontechnical users to build refine coach edit update their agent

[译文] [Clay Bavor]: 因此，我们为客户带来了我们称为 Agent OS 的平台，这是我们构建智能体的平台。它是一套非常复杂的工具包，用于通过代码构建非常复杂的、面向客户的智能体。回到“架构师”这一点，构建卓越的面向客户的智能体还有另一面，那就是体验、品牌和市场。因此，与之配套的是，我们有一套无代码工具，使非技术用户能够构建、完善、指导、编辑和更新他们的智能体。

[原文] [Clay Bavor]: and importantly these two seamlessly interoperate and and so I think when folks approach us and say hey we're just going to roll our own they look at oh my goodness all of the things under the surface of the iceberg the problems that we have spent you know the better part of two years uh running into and then solving and pulling together in a very coherent platform to build these scaled customerf facing agents that can pick up the chat pick up the phone um and handle a high degree of complexity and the set of tools for non-technical users to contribute uh to to the agent as well

[译文] [Clay Bavor]: 重要的是，这两者可以无缝互操作。所以我想，当人们找上门来说“嘿，我们要自己搞”时，他们随后会看到冰山水面下的所有东西——那些我们花了两年多时间遇到并解决的问题，我们将解决方案整合进了一个非常连贯的平台中，用于构建这些大规模的、面向客户的智能体。这些智能体可以接听聊天、接听电话并处理高度复杂的情况，同时这套工具也让非技术用户能够为智能体做出贡献。

[原文] [Clay Bavor]: and that rings quite true and where we have had um companies we've interact with go down the path of build your own we've had many of them come back nine months later and it's like hey uh it was deeper and darker than we expected under there you know can we talk so uh that's kind of the journey and the the pattern that we see

[译文] [Clay Bavor]: 这被证明是非常真实的。我们与之接触过的一些公司走上了“自建”的道路，但许多公司在九个月后又回来了，说：“嘿，下面的情况比我们预期的更深、更黑，我们能谈谈吗？”这就是我们看到的过程和模式。

章节 5：开发新范式——智能体生命周期与仿真测试

📝 本节摘要：

针对 AI 智能体这种“非确定性软件”，Clay Bavor 提出了一套全新的开发流程——“智能体开发生命周期（Agent Development Life Cycle）”。他指出解决 AI 问题的核心在于“用更多的 AI 去测试 AI”。Sierra 构建了一套用户模拟测试工具，能生成数十种不同的人设和场景（如设备故障）对智能体进行压力测试。在上线前，智能体会经历数万次模拟对话以发现盲点。上线后，通过闭环工具分析人工接管（Hand-off）案例，让智能体从错误中学习，实现性能的“螺旋式上升”。

[原文] [Alessio Fanelli]: yeah what's the agent building iteration process like when people are building on Sierra or when you're seeing people build agents like how should people think about how to push the envelope and you can also do things like you couldn't AB test a customer support person before now with Sierra you can kind of have different personalities like do you see people be very creative with that

[译文] [Alessio Fanelli]: 是的，当人们在 Sierra 上构建时，或者当你看到人们构建智能体时，智能体的构建迭代过程是怎样的？人们应该如何思考如何突破极限？而且你还可以做一些以前做不到的事情，比如你以前无法对客服人员进行 A/B 测试，现在有了 Sierra，你可以拥有不同的个性。你是否看到人们在这一点上非常有创意？

[原文] [Clay Bavor]: it's a couple levels one uh we've had to essentially event invent a new software development life cycle we call it the agent development life cycle where you have this non-deterministic piece of software So how do you test it well one of the things we've discovered is the solution to most problems with AI is more AI and so when you're testing a company's agent how do you do that you can't just put in a single input and hope you get the the right output

[译文] [Clay Bavor]: 这包含几个层面。首先，我们本质上不得不发明一种新的软件开发生命周期，我们称之为“智能体开发生命周期”，因为你面对的是这种非确定性的软件。那么你如何测试它呢？我们发现的一件事是，解决 AI 大部分问题的方法是使用更多的 AI。所以当你测试一家公司的智能体时，你该怎么做？你不能只是输入一个单一的输入并希望得到正确的输出。

[原文] [Clay Bavor]: we've built a whole user simulation testing harness where we can create dozens of different personas with simulated accounts uh even simulated devices that they're troubleshooting and you know the amber light is on or off and and so on and so first and foremost have had to think through all of the parts of the software development life cycle

[译文] [Clay Bavor]: 我们构建了一整套用户模拟测试工具（Harness），我们可以在其中创建数十种不同的用户画像，带有模拟账户，甚至他们正在排查故障的模拟设备——比如黄灯是亮着还是灭着等等。所以首要的是，必须想清楚软件开发生命周期的所有部分。

[原文] [Clay Bavor]: um with that as the foundation you then have this approach to building out every every business's agent which starts with deeply understanding what they're trying to do with their customers what are the key customer journeys and then we have a variety of techniques for modeling those in code uh in a way that is very expressive uh and lets the agent simultaneously kind of hit the curve balls and flex if someone comes in on one topic and goes to another do that but then when it matters right be you know down to fully deterministic where needed like there's no hallucinating you know compliance regula uh compliance language that that you want

[译文] [Clay Bavor]: 以此为基础，接下来是构建每家企业智能体的方法。首先要深入理解他们试图与客户做什么，什么是关键的客户旅程。然后我们有各种技术用代码对这些进行建模，以一种非常有表现力的方式，让智能体能够同时应对突发状况并灵活变通——如果有人从一个话题切入然后转到另一个话题，它可以处理。但在关键时刻，它必须是完全确定性的，比如在需要的时候，不能在合规语言上出现幻觉。

[原文] [Clay Bavor]: from there we we then use uh the uh simulations testing harness to in essence have tens or hundreds of thousands of conversations with the agent before it's live for the first time and from there we can tell oh you know it doesn't know enough about this part or it needs to be able to handle this corner case better and so on

[译文] [Clay Bavor]: 从那时起，我们使用模拟测试工具，在智能体首次上线之前，本质上与它进行数万甚至数十万次的对话。从中我们可以判断：“哦，它对这部分了解不够，或者它需要更好地处理这个极端情况（Corner Case）”等等。

[原文] [Clay Bavor]: and then um it it really gets interesting when we go live and we have a a set of tools that give CX teams and engineering teams deep insight into where does the agent realize like oh like I'm I'm beyond my abilities on this i'm going to hand off to to a person uh and then we have this closed loop uh set of tools where the agent can learn from its past mistakes it can be coached it can be improved and you end up with this kind of upward spiral of performance and capability

[译文] [Clay Bavor]: 然后，当我们上线时事情变得真正有趣起来。我们有一套工具，能让客户体验（CX）团队和工程团队深入洞察智能体在哪里意识到：“哦，这件事超出了我的能力范围，我要转接给人工。”然后我们有这套闭环工具，智能体可以从过去的错误中学习，可以被指导，可以被改进，最终你会得到这种性能和能力的螺旋式上升。

章节 6：驾驭技术浪潮——关注“一阶导数”与预判未来

📝 本节摘要：

面对 AI 技术令人眼花缭乱的迭代速度，Clay Bavor 分享了他保持敏锐的秘诀。他认为，除了常规的阅读论文和社交媒体资讯外，最重要的是“亲自上手（Hands-on）”体验技术，甚至是目前尚未直接使用的相邻技术（如视频模型）。他提出了一个核心思维模型：关注“一阶导数”（变化率）而非绝对状态。这意味着产品决策应基于对未来的预判（如预测 Token 成本会暴跌），也就是“滑向冰球将要去的地方”。他甚至保留了一份 Google 文档，专门记录那些 GPT-4 无法解决的难题，并定期用新模型（如 o1, o3）进行测试，以精准描绘技术进步的斜率。

[原文] [Alessio Fanelli]: yeah you mentioned beyond my ability what's your process for like staying up to date on the ability of the models i think there's a lot of people that try a model or try an app and it's like it doesn't work but then it works a month later because the models improve so quickly uh what what's your process for staying up to date on it

[译文] [Alessio Fanelli]: 是的，你提到了“超出我的能力范围”。你了解模型能力的流程是怎样的？我想很多人尝试了一个模型或一个应用，发现它不管用，但一个月后它又管用了，因为模型进步得太快了。你保持跟进的流程是什么？

[原文] [Clay Bavor]: well first of all if you feel like things are changing faster than they ever have before it's because they are um I I I feel like whether it's I don't know Dance Dance Revolution or Beat Saber just like new models new agent frameworks uh uh new benchmarks and so on are just coming down the pike at at an incredible and increasingly fast rate

[译文] [Clay Bavor]: 首先，如果你觉得现在的变化速度比以往任何时候都快，那是因为事实确实如此。我觉得无论是像《劲舞革命（Dance Dance Revolution）》还是《光剑节奏（Beat Saber）》那样，新模型、新智能体框架、新基准测试等等，正以令人难以置信且日益加快的速度接踵而至。

[原文] [Clay Bavor]: so I think one thing is I think fairly typical I think we all do dipping into Twitter acts uh reading the latest research papers and and really trying to just immerse oneself in things that are even adjacent to what specifically you're doing so we don't yet use video models but gosh what the you know V3 models are cap capable of and uh what's emerging in like okay it gives it gives you a hint of what's going to be around the corner maybe in the area that that you're directly uh working in

[译文] [Clay Bavor]: 所以我认为有一件事是相当典型的，我想我们都在做：浏览 Twitter/X，阅读最新的研究论文，并真正尝试让自己沉浸在即使只是与你具体工作相邻的事物中。比如我们还没使用视频模型，但天哪，看看 V3 模型能做什么，以及正在涌现什么。这会给你一个暗示，告诉你什么即将到来，也许就在你直接工作的领域里。

[原文] [Clay Bavor]: and then I just think there is no substitute for hands-on and using it and so uh really being handson with the tools whether or not right again you're directly applying them in in your work or what it is you're I think is so important and I I would argue that uh understanding where things are going is even more important than understanding where things are today so the first derivative is more important than kind of the absolute uh state of capability

[译文] [Clay Bavor]: 其次，我认为没有什么能替代“亲自上手”和实际使用。所以，真正动手使用这些工具——无论你是否直接将其应用在工作中——我认为这都至关重要。我甚至认为，理解事物的发展方向比理解事物的现状更重要。所以，“一阶导数（变化率）”比能力的绝对状态更重要。

[原文] [Clay Bavor]: an example of a decision we made early on was like we had a strong sense when we started the company that the uh cost per token was going to plummet that model capabilities were going to expand and and so you want to be building to where the puck is going as opposed to where it currently is

[译文] [Clay Bavor]: 我们早期做出的一个决定就是一个例子：我们在创办公司时就有强烈的预感，Token 的成本会暴跌，模型能力会扩展。所以你要朝着“冰球将要去的地方”去构建产品，而不是它现在所在的地方。

[原文] [Clay Bavor]: and and so almost having a ritual where on a on a cadence you're checking in with the capabilities of the model I keep in a Google doc some problems that were too hard right for GBT T4 to solve but you know 01 or 03 because can 03 mini do it um and uh you're you're checking in on the capabilities of these models to basically plot again the slope so that you can understand cool if we start building this now this will intercept you know at this period of time and we could have this level of model capability but with this latency and I think that's that's how you build truly great products that are at the frontier it's by anticipating the frontier

[译文] [Clay Bavor]: 因此，几乎要形成一种仪式，即按照一定的节奏去检查模型的能力。我在一个 Google 文档里保留了一些对 GPT-4 来说太难解决的问题，但是你知道 o1 或 o3……因为 o3 mini 能解决吗？你需要检查这些模型的能力，基本上是为了再次绘制出那个斜率。这样你就能明白：酷，如果我们现在开始构建这个，它将在那个时间段通过（拐点），我们可以拥有这种水平的模型能力，但伴随着这种延迟。我认为这就是你如何构建处于前沿的真正伟大的产品——通过预判前沿。

章节 7：下一代交互——多模态智能体与可穿戴设备

📝 本节摘要：

在这最后的总结章节中，Clay Bavor 展望了 AI 交互的终极形态。他批评目前的交互界面（像 AOL 即时通一样的聊天框）过于原始。他提出了“变形金刚（Shape-shifters）”的概念，即智能体应能根据情境动态调用文本、语音、视频和视觉界面。基于他在谷歌 10 年的 AR/VR 经验，他预言智能眼镜和可穿戴设备将是 AI 的“终极载体”。他认为，为了实现一个“看见你所见、听见你所听”的无所不在的助手，我们必须消除从口袋里掏出“长方形玻璃块”（手机）的摩擦力，让 AI 真正成为人类感官的自然延伸。

[原文] [Alessio Fanelli]: yep um I know we only got a minute and a half left but you spent 18 years at Google you kind of started the ARVR projects you started the lens project what What do you think about the next interface for AI so we had text now we have voice obviously video coming do you see the glasses are going to do you think the glasses are going to work is it more ambient agents like any thoughts

[译文] [Alessio Fanelli]: 是的，我知道我们只剩一分半钟了，但你在谷歌待了 18 年，你某种程度上开启了 AR/VR 项目，开启了 Lens 项目。你怎么看 AI 的下一代交互界面？我们要么是文本，现在有了语音，显然视频也快来了。你认为眼镜会……你认为眼镜会成功吗？它是更多环境感知的智能体吗？有什么想法？

[原文] [Clay Bavor]: yeah I first of all I think how we interact with AI and agents is going to look super different from today today it's like mushed into what looks like AOL Instant Messenger right a chat a chat interface or it's a voice call

[译文] [Clay Bavor]: 是的，首先我认为我们与 AI 和智能体交互的方式将与今天截然不同。今天它就像被塞进了像 AOL 即时通（AIM）那样的东西里，对吧？一个聊天、聊天界面，或者是一个语音通话。

[原文] [Clay Bavor]: i think agents are going to look like shape shifters that can summon text voice video imagery user interface and more and you're going to interact with uh every uh sense and mechanism that that you have

[译文] [Clay Bavor]: 我认为智能体看起来会像“变形金刚（Shape-shifters）”，它们可以召唤文本、语音、视频、图像、用户界面等等，而你将用你拥有的每一种感官和机制与之互动。

[原文] [Clay Bavor]: as for the hardware uh look I spent 10 years of my life uh building uh in AR and VR my strong view is that uh glasses and wearables will be the ultimate vehicle for the trusted personal AI that is with you

[译文] [Clay Bavor]: 至于硬件，听着，我生命中有 10 年时间在构建 AR 和 VR。我的强烈观点是，眼镜和可穿戴设备将是那个伴随你的、值得信赖的个人 AI 的终极载体。

[原文] [Clay Bavor]: something that can see what you see hear what you hear uh that can whisper in your ear or you know nudge you that way uh visually

[译文] [Clay Bavor]: 它可以看见你所见，听见你所听，可以在你耳边低语，或者你知道，在视觉上指引你。

[原文] [Clay Bavor]: i I just think we're on this path to every one of us having an omniresent uh omniapable uh AI assistant that can help us navigate the world uh lead better and healthier lives uh be smarter than we are on our own

[译文] [Clay Bavor]: 我只是认为我们正走在这样一条路上：我们每个人都将拥有一个无所不在、无所不能的 AI 助手，它可以帮助我们在这个世界导航，过上更好、更健康的生活，让我们比独自一人时更聪明。

[原文] [Clay Bavor]: and I think you know going into your pocket or purse or bag to retrieve the you know rectangle of glass and metal and you know swipe up and uh whatever it is I I just I think for such an important capability that will feel in time like an extension of ourselves you you want that to be with you throughout the day and so I think wearables I think glasses will be a a central part of that and uh it's something I'm I'm super excited to see emerge

[译文] [Clay Bavor]: 我认为，把手伸进口袋、钱包或包里，取出那块玻璃和金属构成的长方形，然后向上滑动，无论是什么操作——我认为对于这样一种随着时间推移感觉就像我们自身延伸的重要能力来说，你会希望它整天都伴随着你。所以我认为可穿戴设备，我认为眼镜将是其中的核心部分，这是我非常兴奋能看到其涌现的事物。

[原文] [Alessio Fanelli]: awesome thank you Clay thank you so much

[译文] [Alessio Fanelli]: 太棒了，谢谢你 Clay，非常感谢。

[原文] [Clay Bavor]: thank you thanks everyone

[译文] [Clay Bavor]: 谢谢，谢谢大家。