章节 1：开场与嘉宾背景：从“一日律师”到营销专家的职业转型

📝 本节摘要：

本节作为访谈的开篇，主持人 Jonathan 热情介绍了嘉宾 Steve Cunningham，回顾了 Steve 曾作为其 AI 启蒙导师的渊源。Steve 分享了自己极具戏剧性的职业背景：他曾是一名法学院毕业生，却在正式成为律师仅一周后便辞职，原因是他发现律师的工作枯燥乏味，且大多数合伙人的生活状态并非他所向往。出于对创造力的追求和对商业书籍的兴趣，他最终转型创立了自己的营销代理公司。

[原文] [Jonathan]: we are going to go over my setup for how I am using specifically cloud co-work which is a layer on top of cloud code and by the time this comes out I'm assuming all of the other large language models will have some version of this watch but basically it's like the promise of agents is here It's in tools like cloud co-work and there's a lot of things you have to do right in order to leverage it well and I want to show my setup and what I want to show specifically is how it's evolved even in just in 3 weeks So gtming good afternoon everyone It is I Jonathan Carford aka Coach K This is another episode of the new season of 2026 GTM AI podcast and I am actually super geeked if that's the right term to have my man Steve Cunningham on with me When I first started my AI journey I started a company who helped with AI and I was looking for any education I could around AI besides my own company And Steve's company was one of the one I went through his program three years ago three and a half four years It's been a while now And so he was one of my um I guess you say one of my AI mentors along the way which I'm super grateful and just incredibly humbled you'd be willing to come back on and talk to me about this since you are a part of my journey along with this AI guy in India who helped me learn a lot too So just I just want to tell you first off thank you for being a part of the journey I'm just super excited to have you How are you

[译文] [Jonathan]: 我们今天要以此为主题，探讨我是如何具体使用 Claude Cowork 的，这是构建在 Claude Code 之上的一个层级。等到这期节目播出时，我假设所有其他的大型语言模型都会推出类似的功能供大家观看。但这基本上意味着 Agent（智能体）的承诺已经兑现，就在像 Claude Cowork 这样的工具中。为了用好它，你必须做对很多事情，所以我想展示我的设置，特别想展示它是如何在短短 3 周内进化的。那么，GTM 正在进行中，大家下午好。我是 Jonathan Carford，人称 Coach K。这是 2026 GTM AI 播客新一季的又一集，我现在真的超级——如果这个词恰当的话——“极客般兴奋”（geeked），能邀请到我的老兄 Steve Cunningham。当我最初开始我的 AI 之旅时，我创办了一家帮助应用 AI 的公司，除了我自己的公司外，我当时在寻找任何能找到的 AI 教育资源。Steve 的公司就是其中之一，我大概在三年前，或者三岁半、四年前参加过他的课程。已经有一段时间了。所以他算是我的一位……我想你可以说是我一路走来的 AI 导师之一，对此我超级感激，而且你愿意回来和我聊这个话题让我感到非常荣幸，因为你和另一位帮助我学到很多的印度 AI 专家一样，都是我旅程的一部分。所以我只想首先对你说，谢谢你成为这段旅程的一部分。我真的很兴奋能邀请到你。你好吗？

[原文] [Steve]: Yeah I'm doing really well I don't know if it was three or four years ago Probably three and yeah I'm just super pumped for what you built and since we last spoke So great to be here

[译文] [Steve]: 耶，我过得非常好。我不确定那是三年前还是四年前，可能大概是三年前吧。是的，我也为你所建立的一切感到超级兴奋，尤其是自从我们上次交谈以来。所以很高兴来到这里。

[原文] [Jonathan]: Yeah it's been a blast So you obviously when I joined you you were running the Simple Academy and you've had read it for me You've been involved with tons of stuff since But your launch by the time this comes out you would have launched humans plus agents which I'm super excited to hear about But before we get into the business side of it I'd love to know my audience always loves to know the human behind who the heck we're talking to So tell me about Steve who is the non-LinkedIn non-res version Who the heck is Steve and how did you get this crazy AI shenanigans

[译文] [Jonathan]: 是啊，这真是一段很棒的经历。显然当我加入你的课程时，你正在运营 Simple Academy，而且你有 ReadItForMe（读书摘要服务）。从那以后你涉足了很多事情。但这期节目播出时，你应该已经推出了“Humans + Agents”（人类+智能体）项目，我超级期待听到关于它的内容。但在我们进入商业话题之前，我想了解一下——我的听众总是喜欢了解我们交谈对象背后的“人”。所以跟我说说 Steve，那个非 LinkedIn、非简历版本的 Steve 是谁？Steve 到底是谁？你是怎么卷入这场疯狂的 AI 恶作剧（shenanigans）中的？

[原文] [Steve]: Yeah Well I always start off my story by telling everybody I was a lawyer for exactly one week which is a true story Grew up in just outside of Toronto Canada What else can I tell you before the work stuff played hockey as a kid growing up Canadian boy Played college hockey Went to school in Maine at Kohley College Did a lawyer thing for one week and then created a marketing agency Did that for about 10 years

[译文] [Steve]: 好的。我也总是以告诉大家我当了“整整一周”的律师来开始我的故事，这是个真事。我在加拿大多伦多郊外长大。在工作之外还能告诉你什么呢？作为一个加拿大男孩，小时候打冰球，大学也打冰球。我在缅因州的 Colby College 上学。做了一周的律师，然后创建了一家营销代理公司。在那行干了大概 10 年。

[原文] [Jonathan]: So hang on a second when you were a lawyer like what made you go I'm going to do marketing What was the thing that made you go this is not for me What was that

[译文] [Jonathan]: 等一下，当你是个律师的时候，是什么让你决定“我要去做营销”？是什么让你觉得“这行不适合我”？那是什么原因？

[原文] [Steve]: Well so the story is in Canada when you're in law school you get a job in the summer usually between your second and third year and what you'll usually do is go basically carry the bag as a senior partner and just go shadow them and help them and kind of get a chance to see what it all be about And I realized that that was going to be my life They don't tell you this before a lot Like my advice to anybody who's thinking of doing a profession is to like to go do the work Well I love law school was amazing super intellectually stimulating all the good things But the work of being a lawyer seemed to me at that time to be very mind-numbing And most of the partners that I met hated their lives Not all of them Some of them were engaged in their practice and all that good stuff But I just decided that was going to be my life and I'd rather try to make my own way as an entrepreneur and that was it And the story with the one week part is in Canada they called articling which is basically like a internship for your a full lawyer and as I was finishing up my articling they asked me to come back as a lawyer and I said no I don't want to I'm leaving and they said well you're working on this one file could you just stick around to finish finish it finish the file So I said "Yeah great Then I'll have a story to tell for the rest of my life and I'm going to go." So yeah that's that was why I was a lawyer for exactly one week

[译文] [Steve]: 那个故事是这样的，在加拿大读法学院时，你通常会在第二年和第三年之间的暑假找份工作。你通常做的基本上就是给高级合伙人“拎包”（carry the bag），去跟随他们，给他们打下手，这让你有机会看看这一行到底是怎么回事。然后我意识到那将是我未来的生活。在此之前没人告诉你这些。我对任何想从事某个职业的人的建议是，去实际做那个工作。我喜欢法学院，那很棒，智力上非常受启发，全是好东西。但律师的实际工作在当时对我来说似乎非常令人麻木（mind-numbing）。而且我遇到的大多数合伙人都讨厌他们的生活。不是所有人，有些人投入在他们的业务中，也不错。但我就是认定那将是我的生活，而我宁愿尝试作为创业者走出自己的路，就是这样。关于“一周”的故事是，在加拿大有一种叫“Articling”（见习期）的制度，基本上就像是正式成为律师前的实习。当我要结束见习期时，他们邀请我回来做正式律师，我说不，我不想，我要走了。他们说：“好吧，你正在处理这一个案卷，你能留下来把它做完吗？把案卷结了。”所以我说：“好啊，太棒了。这样我就有一个可以说一辈子的故事了，然后我就走人。”所以是的，这就是为什么我当了整整一周的律师。

[原文] [Jonathan]: And why So why marketing Because of all the things you could have done from a lawyer Why marketing

[译文] [Jonathan]: 那为什么？为什么是营销？因为从律师转行你可以做很多事，为什么选营销？

[原文] [Steve]: Well I had read a whole bunch of business books because I had a my family had a very small signage company and I one of the things I wanted to do was to join the family business So I'd read a whole bunch of marketing and the people or the business book authors who seemed the most interesting to me were the marketing people because they were like crazy ideas and offthe-wall kind of stuff right So it was I put no more thought into it than that It seemed like that kind of work would be fun I've always been somebody who was creative and liked to create things and so marketing was a you know you get to go to make stuff up all day and thought it would be fun It turns out I didn't have that much fun running a marketing agency but you know I you got to learn these lessons more than once before they stick

[译文] [Steve]: 我读了一大堆商业书籍，因为我家里有一个非常小的标志牌公司，我想做的事情之一就是加入家族企业。所以我读了一大堆营销书，而在那些商业书作者中，我觉得最有趣的就是营销人员，因为他们总是有疯狂的点子和稀奇古怪的东西，对吧？所以我其实没多想其他的。看起来那种工作会很有趣。我一直是一个有创造力、喜欢创造事物的人，而营销嘛，你知道，你可以整天去编造东西，我觉得那会很好玩。事实证明，经营一家营销代理公司并没有那么好玩，但你知道，有些教训你得学个不止一次才能长记性。

[原文] [Jonathan]: Well I can say from being on the other side of it at some part of your path like it seems like you seem to have mastered the marketing world And obviously with AI you could tell me better than anyone that it's definitely shifted a ton since when you started marketing So just seeing yourself who's been in the marketing space and then adjusting as you need to and still having success I think is really cool So first off congrats on that I think it's great

[译文] [Jonathan]: 我可以说，从另一方面来看，在你职业路径的某个阶段，似乎你已经掌握了营销界。显然随着 AI 的出现，你比任何人都更清楚，这与你刚开始做营销时相比已经发生了巨大的转变。所以看到你在营销领域深耕，然后根据需要进行调整，并且依然取得成功，我觉得真的很酷。所以首先，恭喜你，我觉得这很棒。

章节 2：AI 浪潮下的觉醒：商业模式的颠覆与“人类+智能体”的诞生

📝 本节摘要：

在本节中，Steve 坦诚地分享了 AI 如何在一夜之间颠覆了他经营十年的读书摘要公司 ReadItForMe。他回忆了面对技术冲击时的复杂心情——如果可以，他甚至想回到过去摧毁 AI。但现实迫使他迅速转型，从单纯的“读书人”转变为帮助企业应用 AI 的先行者。对话还探讨了过去三周内 AI 技术的巨大飞跃，以及 Steve 提出的全新商业理念：“Humans + Agents（人类+智能体）”经济，这也为接下来的实操演示奠定了理论基础。

[原文] [Jonathan]: Thank you Yeah that's really cool So then AI so how where was the transition from marketing to AI What was the thing Was it when chatbuchi launched and was going crazy and same thing What was the aha moment for you

[译文] [Jonathan]: 谢谢，是的，这真的很酷。那么说到 AI，你是如何从营销转型到 AI 的？转折点在哪里？是因为 ChatGPT（注：原文误识别为 chatbuchi）发布并引发疯狂的时候吗？对你来说，那个“顿悟时刻”（aha moment）是什么？

[原文] [Steve]: Yeah so I ran a company called read for me for about 10 years before I read this was my I always like to joke and I still think this is true True If I could take a time machine back to November 2022 and destroy all the AI I would still do it because that job which allowed me to read books for a living and share what I was learning with people like as a lifelong learner That was like the best I could imagine myself doing that for the rest of my life I was not super pumped when AI came out and I realized very quickly that I could get a decent book summary in about five minutes where in the past it would have taken me eight hours to read the book and make notes and be super detail oriented about it And I was really good at it Like I was probably it's a weird thing to be the best in the world at but I was probably the best at reading business books and summarizing them in a really simple understandable way And this AI thing almost immediately was as good or better than I was And so it didn't take me long to figure out that if I could figure out how to do that everybody could figure out how to do that And once everybody figured out how to do it the business would be very easy to replace Like I AI gobbled that business up pretty quickly And it wasn't it wasn't like overnight the like the business tanked but slowly but surely people started figuring out that they could do whatever they wanted with AI and learn anything that they want with AI including getting summaries of really good business books So that was when we basically got into the space And so I was both embedding AI into our business model to trying to ramp up our production And when we were doing that a lot of people wanted to know how we were doing it which is I think is was around the time when we got connected So we just launched a a program and that was kind of the jumping off point into all the stuff that we for the last three years

[译文] [Steve]: 是的，在这一切发生之前，我经营了一家名为 ReadItForMe 的公司大约 10 年。我总是喜欢开玩笑——而且我现在仍然认为是真的——如果我能坐时光机回到 2022 年 11 月，摧毁所有的 AI，我还是会那么做。因为那份工作让我可以靠读书为生，并作为一个终身学习者分享我所学到的东西。那是我能想象到的最好的工作，我愿意余生都做这个。所以当 AI 出现时，我并没有特别兴奋。我很快意识到，我可以在大约五分钟内得到一份像样的书摘，而在过去，这需要我花八个小时去读那本书、做笔记，并且非常注重细节。而且我真的很擅长这个。虽然说自己是“世界上最擅长读商业书并将其简化为易懂摘要的人”听起来有点怪，但我当时可能确实是最好的。而这个 AI 几乎立刻就做得跟我一样好，甚至更好。所以我很快就明白，如果我能搞清楚怎么用 AI 做这个，每个人都能搞清楚。一旦大家都知道了，这门生意就很容易被取代。就像是，AI 很快就吞噬了那项业务。它并不是一夜之间崩溃的，但缓慢而确定的是，人们开始意识到他们可以用 AI 做任何想做的事，学任何想学的东西，包括获取优质商业书籍的摘要。所以那就是我们基本上进入这个领域的时候。当时我一方面将 AI 嵌入我们的商业模式以试图提高产量，当我们这样做时，很多人想知道我们是怎么做到的，我想那大概就是我们就取得联系的时候。所以我们当时刚推出了一个项目，那就是我们要讲的过去三年所有事情的起跳点。

[原文] [Jonathan]: You know you know what's fascinating is I think about those times because when I was doing this stuff with you was in the spring of 23 So I had long story I don't know if I ever told you this but I was at a company where the company had bought an AI technology in 21 and the guy built his own model from India He came on to be a chief AI officer and I learned all this stuff because I was responsible for training the team and anyways he told me about chatbt launching out and then I was like I got to find other people who are doing stuff in this because I want to make sure I know because he wasn't a marketing guy he was an AI guy So I found your stuff And I was like "This is perfect." So that was back then in G G G G G G G G G G G G G G G G G G G G GBT3 3.5 and I remember the prompts we used to make back then versus how you prompt now with like recursive learning and all this stuff it's fascinating but it gave me this foundation of understanding how to learn and seeing you as a again not as an AI guy but as a marketer go through and think through how to do things like oh this is brilliant so I was just it's always been something I've been grateful for and it's cool to see how it's moved since then in three years you know so it's been transformational over the last three years and I'm sure we'll get

[译文] [Jonathan]: 你知道吗，回想那个时候真的很有意思，因为我和你一起搞这些东西是在 2023 年的春天。说来话长，我不确定是否跟你讲过，我当时在一家公司，他们在 2021 年就收购了一项 AI 技术，那个家伙在印度构建了自己的模型。他来担任首席 AI 官，我学到了所有这些东西，因为我负责培训团队。总之，他告诉我 ChatGPT 推出了，然后我就想，我得找找其他在这个领域做事的人，因为我想确保我懂行，毕竟他不是营销人员，他是 AI 技术人员。所以我发现了你的东西，我当时就觉得：“这太完美了。”那还是在 GPT-3、3.5 的时代。我记得我们当时用的提示词（prompts），对比现在用的递归学习（recursive learning）之类的东西，这真的很迷人。但这给了我一个理解如何学习的基础，看到你——再次强调，不是作为一个 AI 技术人员，而是作为一个营销人员——去经历并思考如何做事，我觉得这太聪明了。所以我一直对此心存感激，看到这三年来的变化真的很酷，你知道吗，这三年是变革性的，我相信我们会谈到这个。

[原文] [Steve]: And in the last 3 weeks it has been like the biggest shift right since the whole thing started like 0ero to1 no AI to AI was obviously a huge deal and there's been some you know evolutions since then that have been really good for business but in the last three weeks things have starting to really take off and this year is going to be very exciting year for a lot of people and also very scary for a lot of people So you're transitioning from simple academy to human plus agents So tell me about what's the shift what's happening at this room

[译文] [Steve]: 而且在过去的 3 周里，发生了自整件事开始以来最大的转变。从 0 到 1，从无 AI 到有 AI 显然是一件大事，自那以后也有一些对商业非常有益的演进，但在过去三周，事情真的开始起飞了。今年对很多人来说将是非常激动人心的一年，对很多人来说也将是非常可怕的一年。

[Jonathan]: 所以你正在从 Simple Academy 转型到 Humans + Agents（人类+智能体）。跟我说说这个转变，这个领域正在发生什么？

[原文] [Steve]: Yeah it's we've been working with companies over the last three years to help their employees adopt AI And yeah when we started off it was like an academy It's like we're teaching people how to do these things And more and more over those last three years it's all been about helping companies generate better business results So it's not like we're teaching people we're building we built software in the background We built the tools to help companies adopt AI We've got agents that are that we have built that are being deployed We do custom agents and stuff So the it just felt like the right time with this big inflection point that we were at to a better expression of who we are And we think you know trying to put a name on it like what kind of world are we going to be living in It's the humans plus agents economy and figuring out how to humans and agents work together is going to be like job number one for companies over the next two years And so we're trying to position ourselves well And and as a marketing guy we like to change things up every once in a while

[译文] [Steve]: 是的，过去三年我们一直在与公司合作，帮助他们的员工应用 AI。刚开始时，它就像一个学院（Academy），我们在教人们如何做这些事。但在过去三年里，重心越来越变成了帮助公司产生更好的业务结果。所以不仅仅是教人，我们在后台构建了软件，构建了帮助公司应用 AI 的工具。我们要么部署我们要么构建好的 Agent（智能体），要么做定制化的 Agent 之类的。所以感觉现在正是一个合适的时机，在这个巨大的拐点上，更好地表达我们是谁。我们在想，给我们要生活的这个世界起个什么名字呢？这就是“Humans + Agents Economy”（人类+智能体经济）。弄清楚人类和智能体如何协同工作，将是未来两年公司的首要任务。所以我们要努力定位好自己。而且作为一个营销人员，我们也喜欢时不时地搞点变化。

[原文] [Jonathan]: I love it I be creative So the one thing out of this call you told me about before but if someone listens to the end what is what can they expect to get out of listening to you and what you're going to show off today

[译文] [Jonathan]: 我喜欢，这就是创意。关于这次通话有一件事你之前跟我提过，但如果有人听到最后，他们能期待从你的分享中得到什么？你今天打算展示什么？

[原文] [Steve]: We are going to go over my setup for how I am using specifically cloud co-work which is a layer on top of cloud code And by the time this comes out I'm assuming all of the other large language models will have some version of this much but basically it's like the promise of agents is here it's in tools like cloud co-work and there's a lot of things you have to do right in order to leverage it well and I want to show my setup and what I want to show specifically is how it's evolved even in just of three weeks so I have a a backend system that I've got set up that enables me to do what previously would take me weeks to do I get stuff done in 30 minutes and so it's been transformative for our business

[译文] [Steve]: 我们将仔细过一遍我的设置，具体是关于我是如何使用 Claude Cowork 的，这是构建在 Claude Code 之上的一个层级。等到这期节目播出时，我假设所有其他的大型语言模型都会推出类似的版本，但基本上这就是 Agent（智能体）承诺兑现的时刻，就在像 Claude Cowork 这样的工具中。为了用好它，你必须做对很多事情。我想展示我的设置，特别想展示它是如何在短短三周内进化的。所以我建立了一套后端系统，使我能够完成以前需要数周才能完成的工作——现在我 30 分钟就能搞定。这对我们的业务来说是变革性的。

[原文] [Jonathan]: Awesome Hoping some of the things I show here today will help people take a couple of good steps in the right direction And if you listen to the end I'm going to have Steve offered to make sure you guys have a black belt training which is I can I obviously as taking your training I can totally recommend what you do but we'll make sure we give you people away a coupon come in which would be awesome So make sure you stay around and we'll make sure Steve gives that away which should be fun So I'm excited Well want to get into it

[译文] [Jonathan]: 太棒了。希望今天展示的一些东西能帮助人们在正确的方向上迈出几步。如果你听到最后，我会让 Steve 确保给大家提供一个“黑带训练”（black belt training）。显然作为参加过你培训的人，我可以完全推荐你们的内容，但我们会确保给大家发放一个优惠券，这会很棒。所以请务必留下来，我们会确保 Steve 发放那个福利，这应该很有趣。我很兴奋。那么，想开始了吗？

章节 3：实操演示：Claude Cowork 的核心工作流与文件夹架构

📝 本节摘要：

在本节中，访谈进入实操演示环节（Show and Tell）。Steve 共享屏幕展示了他基于 Claude Cowork 的核心工作流。他揭示了自己的文件夹结构在短短三周内如何从“灾难现场”进化为井井有条的“公司化架构”。Steve 引用了 OpenAI 关于 AI 发展的五阶段预测（从聊天机器人到完整的组织），并解释了他如何将 AI 设置视为一家运营中的公司——拥有“高管团队”文件夹和“COO（首席运营官）”文件，以便在启动新任务时让 AI 迅速“入职”并掌握上下文。

[原文] [Jonathan]: Should we show off the goods Let's do it Let's do So are you using cloud co cloud code Oh yeah all the time Yes Well before I get into my setup how do you have it set up What is your how do you think about setting it up

[译文] [Jonathan]: 我们要展示一下干货吗？来吧，来吧。所以你在用 Claude Cowork 和 Claude Code 吗？噢是的，一直在用。是的。那么在进入我的（或者是你的）设置演示之前，你是怎么设置它的？你是如何构思这个设置的？

[原文] [Steve]: Yeah it's um I know setting up could be a broad terms but like I've I started doing cloud code and I realized when cloud code came out because they want to have a more userfriendly version versus an IDE and that kind of stuff It can be intimidating to a lot of people So for Claude I have a section folder in my folders I have a stack of different things based on what we need to do from like brand to my talking to the transcripts like depending what we do I've even had Claude co-work do video editing for me and I have a training document on how to do video Anyways crazy cool things you can do inside of here So my Claude folder is all trained on or categorized in a way so I can reference those things inside of it and just off to the races and get stuff done really quickly which is nice Pretty cool

[译文] [Steve]: 是的，我知道“设置”这个词可能很宽泛。但我开始使用 Claude Code 时意识到——当 Claude Code 刚推出时，因为他们想提供一个比 IDE（集成开发环境）更用户友好的版本——它对很多人来说可能有点吓人。所以对于 Claude，我在我的文件夹里有一个专门的分区，我有一堆根据我们需要做的事情分类的东西，从品牌（Brand）到我的谈话记录（Transcripts），取决于我们在做什么。我甚至让 Claude Cowork 帮我做过视频剪辑，我有一份关于如何做视频的培训文档。总之，你可以在这里面做很多疯狂又酷的事情。所以我的 Claude 文件夹是经过训练或分类的，这样我就可以在里面引用这些东西，然后立马开干（off to the races），非常快速地把事情搞定，这很不错，非常酷。

[原文] [Jonathan]: Cool Yeah So I'll show you my setup Yeah please And and like I said I want to walk you through kind of like the evolution of it over the last couple weeks So this is cloud code work For those of you who are already using it you'll know that if you're not it's a it's an interface on top of cloud code that makes it easy for people like me who are nontechnical to me too What otherwise would be technical things like over the holidays I learned how to use the terminal for the very first time So I need to learn how to use cloud code I'm still I'm getting better at using the terminal and I know that it's actually not that much different than using this but this for some reason feels comfortable because you're used to it

[译文] [Steve]: 酷，是的。那我来展示一下我的设置。是的，请吧。就像我说的，我想带你看看它在过去几周里的进化过程。这就是 Claude Cowork。对于那些已经在使用它的人来说你们懂的；如果你没用过，它就是一个建立在 Claude Code 之上的界面，让像我这样非技术背景的人也能轻松使用。即使是在假期里，我也才第一次学会怎么使用终端（Terminal）。所以我得学怎么用 Claude Code。我在使用终端方面已经越来越熟练了，我也知道它其实和用这个界面没太大区别，但这个界面出于某种原因让人感觉更舒服，因为你习惯了这种方式。

[原文] [Steve]: So anyways one of the one of the things that you that most people know by now I think is that if you give really good context to the AI it can do incredible work If you give it no context you get really bad work What I've done this is my folder It's called operate And essentially over the last 3 weeks I only started use I started using it as soon as it came out So it's been about three days Yeah this is only come out be a month but yeah this is brand new Keep going Sorry So my folder structures evolved quite a bit even in those three days and then this is my I called it my old structure but this like last week and there was a week there was a week before that I didn't really save a snapshot of it and it was even way more messy than this And like this the first week was a disaster The second week you can see I'm trying to create some order here and I'm numbering things and so on

[译文] [Steve]: 总之，我想大多数人现在都知道的一件事是，如果你给 AI 提供非常好的上下文，它就能做出令人难以置信的工作；如果你不给上下文，得到的结果就会很糟。所以我做了这个，这是我的文件夹，名字叫“Operate（运营）”。基本上在过去 3 周里——我一推出就开始用了，所以大概才用了三天——是的，这东西才推出一个月，但这确实是全新的。继续说，抱歉。所以我的文件夹结构甚至在那三天里也进化了不少。这是我称之为“旧结构”的版本，但这其实就是上周的样子。在那之前还有一周，我没保存快照，那简直比这还要乱得多。第一周简直是个灾难。第二周你可以看到我试图在这里建立一些秩序，我开始给东西编号等等。

[原文] [Steve]: This week though one of the one of the things that I always go back to the these thinking lenses that I created and you might remember that from some of the stuff that we did way back when we first started talking but back in the day like back in the day like a few years ago Open AI said "Here's how AI is going to progress over the next 10 years We're going to have chat bots then we're going to have reasoning models Yeah The five we're going to have agents Then we're going to have innovators And then we're going to have organizations and at that time like how the heck is an is AI going to run an organization It does it makes no sense to me But I believe that they're heading in the right direction and we'll keep thinking about it in terms of how might this be an organization So I came back to that thinking last one What if I just set this up like I'm running a company with agents And that's basically how I've been thinking about it for it feels like months but it's been about a week and the it's transformed the way I am able to operate every single day So I have a folder for my executive team My COO file is basically here when I whenever I start a new task it's I say go read the COO folder and get yourself onboarded So I on board the

[译文] [Steve]: 到了这周，我总是回溯到我创造的那些“思维透镜（thinking lenses）”上。你可能还记得我们刚开始交流时做的一些东西。但是在当年——也就是几年前——OpenAI 说：“这就是 AI 在未来 10 年将如何发展：我们将拥有聊天机器人（Chatbots），然后我们将拥有推理模型（Reasoning Models）——是的，那五个阶段——我们将拥有智能体（Agents），然后我们将拥有创新者（Innovators），最后我们将拥有组织（Organizations）。”当时我就想，AI 到底要怎么运营一个组织？这对我来说毫无意义。但我相信他们的方向是正确的，我们会继续思考如果这真的变成一个组织会怎样。所以我回到了那个最后的思考点：如果我就把它设置得像是我在和智能体一起运营一家公司呢？这基本上就是我现在的思考方式，虽然感觉像过了好几个月，但其实才过了一周左右，但这已经彻底改变了我每天的运营方式。所以我有一个“高管团队（Executive Team）”的文件夹。我的“COO（首席运营官）”文件基本上就在这里。每当我开始一个新任务时，我就说：“去读一下 COO 文件夹，然后给你自己做个入职（onboarded）。”所以我让它入职……

章节 4：技术深究（Side Quest）：为何 HTML 是人机协作的最佳语言

📝 本节摘要：

这是一个被称为“支线任务（Side Quest）”的精彩插曲。主持人注意到 Steve 在与他人交流时倾向于使用 HTML 格式而非传统的 Markdown。Steve 对此进行了深度剖析：他认为传统的 Office 文档（Word/PPT）底层代码混乱，不适合大语言模型（LLM）；而程序员常用的 Markdown 虽然对机器友好，但对人类阅读极不友好（缺乏视觉流和图像）。Steve 提出 HTML 是连接人类与 AI 的最佳桥梁——AI（尤其是 Claude）因受过全互联网数据的训练，极擅长编写 HTML，而 HTML 又能呈现完美的视觉效果。这一观点随后得到了 Anthropic（Claude 母公司）最近发布的 Artifacts 功能的验证。

[原文] [Jonathan]: I can pause real fast I saw another comment You talked to someone else about doing an HTML format which I'm curious about because obviously the natural is the MD files right So what makes you do HTML versus MD

[译文] [Jonathan]: 我能不能快速暂停一下，我看到了另一条评论。你跟别人谈到要使用 HTML 格式，我对此很好奇，因为显然通常自然的选择是 MD（Markdown）文件，对吧？所以是什么让你选择 HTML 而不是 MD？

[原文] [Steve]: I'm glad you've asked the okay I try I'll try not to go down too far of a rabbit hole with this but the pre AI the way us knowledge workers would work would be through things like Microsoft office like word documents and powerpoints and excel sheets and so on right those are we don't know any we don't know any different right that's how we started working when we joined the workforce and that those are the tools that we've used ever since and What most people who are deep in the weeds of AI know is like those file types are really bad for LM They can't read them And what most people don't know is that there's so much code wrapped up in those things And so the way I think about it is there's the code layer which is super messy on those programs And then there's the presentation layer It's like what you see on restraint right And that's the one side of it Prei that's how we work So office documents are not good for lls what most software engineers are used to working with markdown files because that's how they do their work and and software engineers are pushing the limits of how these agents work by far the most and so it makes sense that they would be saying we need to use markdown files for these things but markdown files if you ever open one is not really good for understanding as a human it's really hard it's hard to parse there's no images we can't see flow diagram it just doesn't work for humans

[译文] [Steve]: 很高兴你问了这个。好吧，我会尽量不在这上面钻太深的“兔子洞”（跑题太远），但在 AI 时代之前，我们这些脑力工作者的工作方式是通过像 Microsoft Office 这样的东西，比如 Word 文档、PPT 和 Excel 表格等等，对吧？那是……我们不知道别的，我们不知道有什么不同，对吧？那是我们要加入劳动力大军时开始工作的方式，那些也是我们要一直使用的工具。而在 AI 领域钻研很深的人大多知道，这些文件类型对 LLM（大语言模型）来说真的很糟糕。它们读不懂这些文件。大多数人不知道的是，这些东西里包裹着太多的代码。所以我现在的思考方式是，这些程序有一个代码层，那里超级乱；然后还有一个展示层，就像你在屏幕上看到的限制那样，对吧？那是其中的一面。在 AI 之前我们就是这么工作的。所以 Office 文档对 LLM 不好。大多数软件工程师习惯使用 Markdown 文件，因为那是他们工作的方式，而且软件工程师目前是推动这些智能体工作极限的主力军，所以他们说“我们需要用 Markdown 文件做这些事”是有道理的。但是 Markdown 文件，如果你曾打开过一个，你会发现它真的不适合人类理解，真的很难，很难解析，没有图片，我们看不到流程图，它对人类来说就是行不通。

[原文] [Steve]: And so what we need is a language that is both good for large language models and is good for human beings And that language happens to be HTML It works everywhere The agents can read it incredibly well Maybe slightly slower than markdown files but not that much that it makes a difference especially with context windows getting enormously large and the speed at which they can read these things increasing That is the file format that works for very well for humans and it works very well for agents And so we do everything in HTML files and we believe our point of view is that's going to be the thing that unlocks a whole bunch of stuff for everybody The cool thing about it is that what the lens are incredibly good at coding HTML because the entirety of the internet was created on it and that's all that it was trained on What it enables everybody to do is being able to build really cool things with AI and we built a publishing tool to help with this is one of the reasons why we're transitioning into humans plus agents It's like we need to build a world where humans and agents can work well together One of the easiest things to do is to create a file format that works for both

[译文] [Steve]: 所以我们需要的是一种既对大语言模型友好，又对人类友好的语言。而这种语言恰好就是 HTML。它随处可用。智能体能极其出色地阅读它。也许比读 Markdown 文件稍微慢一点点，但并没有慢到产生实质差别的程度，尤其是随着上下文窗口变得巨大，以及它们阅读这些东西的速度在不断增加。这就是那种对人类非常有效，对智能体也非常有效的文件格式。所以我们把一切都做成 HTML 文件，我们的观点是，这将是为所有人解锁一大堆东西的关键。最酷的一点是，LLM 在编写 HTML 方面极其出色，因为整个互联网都是建立在它之上的，而那是它们训练的全部素材。这使每个人都能用 AI 构建非常酷的东西。我们要构建一个发布工具来辅助这一点，这也是我们向“Humans + Agents”转型的原因之一。就像我们需要建立一个人类和智能体能良好协作的世界。最简单的事情之一就是创建一个对两者都适用的文件格式。

[原文] [Steve]: Just today actually it was interesting I've been on this kit for a while and I read a there's a Twitter post and somebody was talking about how nano bananas really great creating images There's all the discussion about PowerPoints are dead Everything's always dead when you read but PowerPoints always the thing that everybody's been like super excited about is we need to do better PowerPoints So I read a thread one time where somebody seemed like they were get I was like why don't we just do HTML I don't get why we need to be building PowerPoints at all You can just build it right in HTML And the conversation went from yeah I can just do those things in HTML And then the punch line was well then I export it out to PowerPoint Well you just did all this work to do in the exact right way and then a lot more work to make it worse for humans and for agents And so just today I'm gonna pull up a couple of tweets here There's a guy from Enthropic who and I feel like this is like the first sign of validation of this idea from somebody at the large language model providers and basically they're obsessed with how we can increase the bandwidth of communication between humans and models or in our language humans and agents and this playground fills you know the jump here Basically what this is the ability for cloud code to build HTML files so that humans can understand them and so it goes into your codebase and it does diagrams that show you how it's architected and all of those kinds of things So that to me is like they are using HTML to do exactly what we're talking about So that's why and I guess time will tell if we're right or not But it's been transformative to the way we do everything in our business It allows me to go through like just an insane number of designer visions of them doing web stuff like custom proposals Like everything is way way easier when you're using H

[译文] [Steve]: 就在今天，实际上这很有趣，我在这件事上已经坚持了一段时间了。我读到一个推特帖子，有人在谈论现在的 AI 生成图像有多棒。总是有关于“PowerPoint 已死”的讨论——当你阅读这些时，好像什么东西总是“死了”，但大家总是对 PowerPoint 超级兴奋，说我们需要做更好的 PowerPoint。有一次我读到一个帖子，我就想，为什么我们不直接用 HTML 呢？我不明白为什么我们非得做 PowerPoint。你可以直接用 HTML 构建它。对话变成了“是啊，我可以直接用 HTML 做那些事”，然后笑点来了：“既然如此，那我再把它导出成 PowerPoint 吧。”好吧，你刚刚费了所有的劲用完全正确的方式做了这件事，然后又费了更多的劲把它变得对人类和智能体都更糟。就在今天，我找几条推特出来。有一个来自 Anthropic（Claude 的开发公司）的人，我觉得这是来自大语言模型提供商的人对这个想法的第一个验证信号。基本上他们痴迷于如何增加人类与模型之间——或者用我们的话说，人类与智能体之间——的沟通带宽。这里有个 Playground 的演示，基本上这就是 Claude Code 构建 HTML 文件的能力，以便人类能够理解它们。它进入你的代码库，画出图表展示架构是如何构成的，诸如此类。这对我来说，就像是他们在用 HTML 做我们正在讨论的确切的事情。这就是原因。我想时间会证明我们是对是错，但这已经彻底改变了我们在业务中做每件事的方式。它允许我快速浏览数量惊人的设计方案，做网页之类的东西，比如定制提案。当你使用 HTML 时，一切都变得简单多了。

[原文] [Jonathan]: Yeah So that's a side quest on that I love it Keep going And sorry and thank you for going down that rabbit hole because I had to ask you I saw you comment because I've used HTML from a design perspective because I can't stand the PowerPoint design tools like all of them gamma like notebook's got cool design but you can't edit anything and anyways I had a team member once in my design team who we were doing some a animations and so we were doing more design with HTML It was so nice This is awesome but we did it for more design What you're doing is moving more information which I love because it kind of combines both worlds So the thought process is invigorating So thank you Keep going Sorry

[译文] [Jonathan]: 是的，所以这是关于那个话题的一个“支线任务”。我喜欢，请继续。抱歉，也谢谢你钻进那个“兔子洞”，因为我必须问你，我看到了你的评论。我也从设计的角度使用过 HTML，因为我受不了 PowerPoint 的设计工具，比如所有的那些，Gamma 之类的，NotebookLM 虽然有很酷的设计但你什么都编辑不了。总之，我设计团队里曾经有个成员，我们当时在做一些动画，所以我们更多地用 HTML 做设计。那感觉太好了。这太棒了，但我们当时是为了做更多设计。而你在做的是移动更多的信息，这我喜欢，因为它某种程度上结合了两个世界。所以这个思维过程令人精神振奋。谢谢你。请继续，抱歉打断。

[原文] [Steve]: Yeah Yeah And so one of one of the bottlenecks then it becomes like publishing these things So like it's very easy to quote unquote publish a PowerPoint but it's hard for most people to publish a web page So I'll just show you one more quick thing that's kind of unlocked a whole bunch of stuff for us which is I just built this really straightforward tool and there and this is really ugly because this is our alpha version of it and it's it's the version that I use but we've turned this into a product know that's part of the humans plus agents platform but being able to just drag and drop like a file and publish it immediately and then boom it's on the internet is like transform it to the workflow if you can get something from the LLM does it you download it you shift it over and it's done and now especially with claude co-work and all the tools that are going to come out because of it like these files I can just drag them and I can send a link to people immediately England people on my team and so on so that's where the like the workflow from beginning to end gets solved with HTML and it's really easy because it's not that hard to publish a web page right It is very hard if you're using a tool like like any of the web platforms like WordPress or whatever Very frustrating But okay the side quest is over Let's go back to how those setup main thing Sorry got the segway but keep going

[译文] [Steve]: 是的，是的。所以接下来的瓶颈之一就变成了发布这些东西。就像，我们要“发布”一个 PowerPoint 很简单，但对大多数人来说发布一个网页很难。所以我再快速展示一个东西，它在某种程度上为我们解锁了一大堆东西。我刚刚构建了这个非常直观的工具，这真的很丑，因为这是我们的 Alpha 版本，是我使用的版本，但我们要把它变成一个产品，作为“Humans + Agents”平台的一部分。能够直接拖放一个文件并立即发布，然后“嘭”的一声它就在互联网上了，这就像是彻底改变了工作流。如果你能从 LLM 那里得到什么东西，它做好了，你下载下来，移过来，就完成了。尤其是现在有了 Claude Cowork 和所有因此将要出现的工具，比如这些文件，我可以把它们拖进去，然后立即发链接给别人，给英国团队的人等等。这就是 HTML 解决从头到尾的工作流的地方，这真的很简单，因为发布一个网页并不难，对吧？如果你用像 WordPress 这样的任何 Web 平台工具，那就很难，非常令人沮丧。好了，支线任务结束了。让我们回到主要的设置上来。抱歉插了一段，但请继续。

章节 5：效能革命：如何用 30 分钟完成价值 1.4 万美元的工作

📝 本节摘要：

在本节中，Steve 展示了他令人咋舌的高效工作流。他将每一项任务都视为一个独立的“项目”，并分享了一个真实案例：他在 30 分钟内完成了一套新平台的用户界面设计，而这项工作如果外包，通常需要 70 个工时、耗资 1.4 万美元。为了量化这种效能提升，Steve 会让 AI 在项目结束时打印一张“工作收据（Receipt）”。此外，他还介绍了自己独特的“日迭代”机制——在每天结束时，要求 AI 复盘并提出系统改进建议，甚至自动更新上下文文档。这意味着他的“AI 公司”每天都在自我进化，每一天的工作量级都相当于传统模式下的一个季度。

[原文] [Steve]: So that's how I get started And then we go into a folder So open up the marketing folder And every one of these folders or like a functional area of my business has a few different things Let's see if I can make this like slightly bigger I don't know if I can Oh that's great It's okay Yeah that's good So every one of these folders has advisors in it So how helping me make better decisions more quickly has agents in it to help me once I decide what I wanted to do the work better context files so that the AI knows exactly what so we're marketing stuff it's going to be like our brand voice design system which is really helpful so you don't have to go back and edit documents it knows exactly what I want to do different templates for like external web stuff internal web stuff and so on and then a project and this is where I do all my work so this these projects is like a if I started a new task I call it a project because we basically are getting like a two to three week type of project done in a 30 minutes

[译文] [Steve]: 这就是我如何开始的。然后我们进入一个文件夹，打开这个“营销”文件夹。每一个这样的文件夹，或者说我业务的每一个职能领域，都有几样不同的东西。我看看能不能把这个稍微放大点……我不知道能不能……噢太棒了，没问题，好的，这样很好。所以，每一个文件夹里都有“顾问（Advisors）”，这是为了帮助我更敏捷地做出更好的决策；里面还有“智能体（Agents）”，一旦我决定了要做什么，它们就帮我把工作做好；还有“上下文文件（Context files）”，这样 AI 就确切知道什么是——比如我们做营销内容，这里就会有我们的品牌声音、设计系统，这非常有帮助，这样你就不必回过头去修改文档，它确切知道我想做什么；还有针对外部网页内容、内部网页内容等的不同“模板（Templates）”；最后是“项目（Project）”，这就是我做所有工作的地方。之所以把新任务称为“项目”，是因为我们基本上是在 30 分钟内完成以往需要两到三周才能完成的那种项目。

[原文] [Steve]: So in my mind we're doing projects So I'll give you an example of a project I did just a couple of days ago as as we were getting ready to launch humans plus agents Got up in the morning this year the first file was at 5:07 So 57 yeah I go to the gym with my family in the morning So I wake up I have a coffee I do an agent I get an agent going So I've always I feel a lot of pressure to have my agents working 24/7 So I got to get them working when I go to the gym just like I showed you here I get it started and I would ask it to do something And here we did a mockup And so basically this is me designing the entire interface for this new platform that we're about to launch And I did about 30 minutes of work total like my time And then the working here for hours at the end One of the things that I do to remind myself of like how much work are we getting done is I I close off every project with a couple of things One is with a I print a receipt for the work that I did So like if I would had to get this done externally how much would this work have cost All the line items it would have it would have been a two to three week project 70 man hours would have cost like $14,000 and I got it done for 30 minutes of my time and with a dollar of computing

[译文] [Steve]: 所以在我的脑海里，我们在做的就是项目。我给你们举个几天前做的项目的例子，当时我们要准备发布“Humans + Agents”。那天早上我起床——今年的第一个文件是在 5:07 创建的，也就是 5 点 07 分，是的，我早上会和家人去健身房。所以我醒来，喝杯咖啡，然后让一个智能体跑起来。我总是感到一种压力，觉得必须让我的智能体 24/7 全天候工作。所以我得在去健身房之前让它们工作起来，就像我刚才展示的那样，我启动它，让它做点事。在这里我们做了一个模型（Mockup）。基本上这就是我为我们即将发布的新平台设计整个界面的过程。我总共只花了大概 30 分钟的时间，就是我自己的时间。然后它们在后台工作了好几个小时。在结束时，为了提醒自己我们到底完成了多少工作，我会做一件事：我在结束每个项目时都会做几步操作。其中之一是，我会打印一张我所做工作的“收据（Receipt）”。就是说，如果我必须找外部人员做这个，这活儿得花多少钱？列出所有细分项目，这本来会是一个两到三周的项目，耗费 70 个工时，大约要花 14,000 美元。而我只用了 30 分钟的个人时间，加上 1 美元的算力成本就搞定了。

[原文] [Steve]: And so one of the one of the things that this enables you to do is to start to understand like the scope of how much work you can actually get done right And if you think in terms of like things that would have taken weeks I can get done in 30 minutes You start to think about how the work gets done which rewires how you plan and how you execute So I treat every day almost like a quarter I treat every task I'm doing like a project and at the end of it I will I'll show you another one that has a wrapup in it Let me find one I did today We're launching another platform called the AI career institute where we train people how to use agents in their specific roles and it's a it's an education one and where did it go There it is So at the end of it I do the invoice but I also do a wrap up And this what this does is I ask the AI at the end of the project or the task in my language project Basically you know make sure you do the invoice So I got the invoice done That's a link to it What do we do So what do we get done What system improvements did we make So one of the things you'll find with these agents is that it'll do something it's not quite right It went off the rails on an instruction So I'll always tell it if I'm in the middle of doing something I find something I don't like I'll say "Go and change that but also go and change the context documents because clearly you didn't understand the instruction Go find out why you didn't understand the instruction and fix that." So it becomes this continuous improvement loop over time

[译文] [Steve]: 这让你能做到的一件事是，开始理解你实际上能完成多大范围的工作，对吧？如果你从“本来需要几周的事情我现在 30 分钟就能搞定”这个角度思考，你就会开始重新思考工作是如何完成的，这会重塑你的计划和执行方式。所以我把每一天几乎都当作一个季度（Quarter）来对待。我把手头的每个任务都当作一个项目。在项目结束时，我会……我给你们看另一个包含“总结（Wrap up）”的项目。让我找一个今天做的。我们正在发布另一个叫“AI Career Institute（AI 职业学院）”的平台，培训人们如何在特定岗位上使用智能体，这是一个教育项目。它哪儿去了？在这儿。所以在结束时，我会做发票（即收据），但我也会做一个总结。这个总结的作用是，我在项目——或者说任务，用我的话叫项目——结束时问 AI，基本上就是：“确保你做了发票。”所以我完成了发票，这是链接。然后问：“我们做了什么？我们完成了什么？我们做了哪些系统改进？”你会发现使用这些智能体时的一件事是，它有时候做的事情不完全对，它偏离了指令。所以我总是告诉它，如果我在做事情的过程中发现我不喜欢的地方，我会说：“去把它改了，但同时也要去修改‘上下文文档’，因为显然你没听懂指令。去找出你为什么没听懂，然后修复它。”所以这就变成了一个随时间推移持续改进的循环。

[原文] [Steve]: And then so in this case we didn't make any system improvements because it was just more like a working session to edit the marketing website for this thing And I didn't get anything wrong in that session And so there was nothing to fix However I also say give me any system improvements that we did that I didn't think about during this And so it gives me a whole list of things that we could make them better So at the end of every day I've got a list of here are all the things we can do to actually make our like our company my company of agents better for tomorrow morning And then I have at the end of the day the I go off and work on that So even overnight the system is getting better So that every day it's like a next version of our company and it's it almost feels like it's a new quarter cuz a whole cycle of projects We've accomplished a whole bunch of work We need to regroup make the whole thing better for the next time And so I wake up in the morning with a better system than I went than I had when I went to sleep So it's that done You see all these projects that these were just the last three days I think those are three days worth of marketing projects and so it is if you get this and if I went back two weeks you would not see that level of productivity and I don't I didn't run an AB test on this So I don't know if this is the reason why I've been able to be so productive over the last week with this but thinking of it in terms of a business and I'm going into a function I'm gonna and I'm gonna do something in that function and fix something in that function has been as I said transformative to the way we do work So the that is the nickel tour through the system that I'm using and I love it

[译文] [Steve]: 那么在这个案例中，我们要修改这个营销网站，这更像是一个工作会话，我们没有做任何系统改进，因为我在那个过程中没发现什么错误，所以没什么可修的。但是，我还说：“给我提出任何我们在这个过程中可能做到、但我没想到的系统改进建议。”于是它给了我一整张清单，列出我们可以改进的地方。所以在每一天结束时，我都有一份清单，上面写着我们能做的所有事情，以便让我们的——哪怕是我的“智能体公司”——在明天早上变得更好。然后在一天结束时，我就去处理那些建议。所以即使在过夜期间，系统也在变好。这样每一天都像是我们公司的一个新版本，感觉几乎就像是一个新的季度，因为经历了一整个项目周期。我们完成了一大堆工作，我们需要重组，让整个系统为下一次做得更好。所以我早上醒来时，拥有的系统比我睡觉前拥有的更好。就是这样。你看所有这些项目，这仅仅是过去三天的量。我想那是三天的营销项目工作量。如果你理解了这个……如果我回溯两周，你看不到这种水平的生产力。我没做 AB 测试，所以我不知道这是否是我过去一周如此高效的唯一原因，但是从“经营一家企业”的角度来思考它——进入一个职能部门，在这个职能里做点事，然后在那个职能里修复点什么——正如我所说，这对我们的工作方式是变革性的。这就是我要展示的系统简明导览（Nickel tour），我非常喜欢它。

章节 6：未来展望：分发、数据与商业本质的不变法则

📝 本节摘要：

在访谈的最终章，话题从工具使用升维至商业战略。Steve 提出，随着技术门槛的降低（例如一夜之间就能克隆出 Instagram），未来的核心竞争力将不再是产品构建能力，而是 分发（Distribution）、客户反馈循环 和 工作流数据。他强调，尽管 AI 能让公司裁掉一半人，但他选择利用释放出的算力去发明全新的价值。面对“未来五年什么不会变”的终极提问，Steve 回归商业本质：客户永远只关心解决问题（止痛药），而非单纯的工具升级（更好的 PPT）。最后，Steve 为听众提供了“黑带训练”福利，Jonathan 以对这位导师的感谢为本次深度对谈画上句号。

[原文] [Steve]: I'm sure it'll be massively outdated by the time this come out But yeah that's how we're thinking of it right now

[译文] [Steve]: 我敢肯定，等到这期节目播出时，这套东西可能已经严重过时了。不过是的，这就是我们目前的思考方式。

[原文] [Jonathan]: What What do you suspect is going to be changing over the next I don't know a couple weeks because like I'm with you like things are dramatically changing very quickly I'm just curious what you see happening

[译文] [Jonathan]: 那么，你推测接下来——我不知道——几周内会发生什么变化？因为我和你一样，觉得事情正在极速剧变。我只是很好奇你预见到会发生什么。

[原文] [Steve]: Well I think the big thing is that that everything is just getting better and that if you know what you want and also you know what good looks like your job your life just gets easier and easier You can you're literally speaking to a computer but getting like I said weeks worth of work done in minutes which requires you to know what you want because a lot of people really get focused on getting it to do cool stuff which is great like you need to explore and you need to understand like what it can do and what it can't do and you know you can have a little fun while you're at it but pointing out things that are productive is going to be like the most important thing So making sure you've got goals in there Making sure you're measuring all the things that the AI is doing Is it moving the needle on the like simple thing like the number of minutes people are spending on your web pages on proposals That's one of that's one of the more interesting ones because I know a lot of go to market folks are in your audience The ability could at scale custom proposals for people and wire it up with Google Analytics to know exactly how long they're spending on the proposal and all that kind of stuff is a really interesting use case for it And we're doing a lot of that with LinkedIn profiles and massively customizing like a super long page you would never even think to try in the past Yeah But in terms of what's going to change like the technology is getting easier to use not harder So it's it's a weird thing where most people think as a technology progresses it gets more and more detailed and technical but it's actually the reverse

[译文] [Steve]: 嗯，我认为最重要的一点是，一切都在变得更好。如果你知道自己想要什么，并且知道什么是“好”的标准，你的工作和生活就会变得越来越轻松。你实际上是在跟一台电脑说话，但正如我所说，能在几分钟内完成几周的工作量。但这要求你必须知道自己想要什么。因为很多人真的太专注于让 AI 做些“很酷”的事情，这当然也没错，你需要去探索，了解它能做什么、不能做什么，过程中也可以找点乐子。但是，指向那些富有成效的事情将是最重要的。所以要确保你在那里设定了目标，确保你在衡量 AI 所做的所有事情。它是否在关键指标上产生了实质影响（moving the needle）？比如人们在你网页上、在提案上花费的分钟数这种简单的事情。这是其中一个比较有趣的用例，因为我知道你的听众里有很多负责市场推广（GTM）的人。能够大规模地为人们定制提案，并将其与 Google Analytics 关联，从而确切知道他们在提案上花了多长时间，诸如此类的事情是一个非常有趣的用例。我们在 LinkedIn 个人资料上也做了很多这样的尝试，进行大规模的深度定制，比如做一个超长的页面，这种事在过去你根本不会想去尝试。是的，但就将会发生什么变化而言，技术正在变得越来越好用，而不是更难。这很奇怪，大多数人认为随着技术进步，它会变得越来越复杂和技术化，但实际上恰恰相反。

[原文] [Jonathan]: I have two questions for you and actually both things I got from you So one thing was I remember when we first went through the thing your spiel was do things better faster easier cheaper with AI sales to pitch and I've realized since then and obviously that's true but there's this dividing line between people doing the old way of things better faster user cheaper and then they're doing completely new things like a whole new world like you said you're thinking in times of instead of three weeks of product you're thinking about doing in a few hours which changes the velocity and the decision-m you make as a result So what do you see yourself doing that's new versus just automating things that were before faster What's the shift for you

[译文] [Jonathan]: 我有两个问题问你，实际上这两点都是我从你那儿学到的。第一点是，我记得当我们最初经历那个阶段时，你的口头禅（spiel）是用 AI 把事情做得“更好、更快、更简单、更便宜”，这是销售话术。从那以后我意识到——显然那也是对的——但在“用老方法把事情做得更好更快更便宜”的人，和“做完全新鲜的事、开启全新世界”的人之间，有一条分界线。就像你说的，你的时间观念从“三周的产品周期”变成了“几小时内完成”，这改变了你的速度以及随之而来的决策方式。那么，你认为自己正在做的哪些事是全新的，而不仅仅是把以前的事自动化加速？对你来说转变是什么？

[原文] [Steve]: So we're in the middle of trying to figure that out right now Like just in the last three weeks I have been able to like I could lay off half of our company tomorrow and it would have no impact and we're not going to It's not the reason why we're doing this But it's also like they don't have work to do I mean the Asians are doing the work now So what do we do So you have to then go invent new ways of adding value So we're working through that and you know part of adding value becomes like it's a research lab right because you don't know what you don't know what's valuable until you try it And using the AI to invent new products and services that you can provide to people that ultimately would enable the engine to keep going is I think the biggest thing

[译文] [Steve]: 我们现在正处于试图弄清楚这一点的过程中。就在过去三周里，我已经能够做到——比如说，我明天就可以裁掉公司一半的人，而这对业务不会有任何影响。当然我们不会这么做，这不是我们做这件事的原因。但也确实存在这样的情况：他们没活儿干了。我的意思是，现在是智能体（Agents，原文误识别为 Asians）在做这些工作。那我们做什么呢？你必须去发明新的增加价值的方法。所以我们正在通过这一点进行探索。你知道，增加价值的一部分变得像是一个实验室，对吧？因为在尝试之前，你不知道什么是真正有价值的。利用 AI 去发明新的产品和服务提供给人们，最终让这个引擎继续运转，我认为这才是最大的事情。

[原文] [Steve]: and like when the way here's the way we think about it There are three things that I think are going to be super important in the future that will be like an actual note First one is distribution um you need to just create more and more distribution because you see all the things on social like you can create I created I coded Instagram overnight so well that yeah you can build anything but if you have zero people to talk to it doesn't matter like the product is not the distribution is one of the things that is worth spending a lot of time on right now so that's number one and then number two you want to get something actually deployed at a company that gets your hooks into them because again it's not the product that is going to be the biggest thing It's going to be getting in the door then finding out what's valuable It's like you're it's like you're in perpetual beta or you know perpetual lean startups And I think listening to what people are struggling with as we make this transition into the humans and agents economy is going to be a skill that is first of all it's only possible if you have distribution because you can only get feedback if you have people that are giving you feedback But because the loop is so tight it's like someone gives me an idea for like a feature that we can add I can have it done overnight and deployed in the morning So yeah but I don't but the way we're thinking about is I don't want to sit around and invent things I want you to tell me what do you want What is going to make your life easier What are you doing right now that you can do that if I solved it for you would be valuable So the frequency at which you can not just ship new things but ship new things that people have asked for and would pay for is another skill that I think is going to be super valuable So we're going to spend a lot of time on that which should create a pipeline of like new stuff for us to do and keep the people busy

[译文] [Steve]: 这里是我们对此的思考方式。我认为未来有三件事将变得极其重要，值得记下来。第一件事是 分发（Distribution）。你需要建立越来越多的分发渠道。因为你在社交媒体上看到了，比如我可以连夜写代码克隆出一个 Instagram，做得非常好。是的，你可以构建任何东西，但如果你连一个说话的对象都没有（即没有用户），那毫无意义。产品不再是重点，分发才是现在值得花大量时间去做的事情之一。那是第一点。第二点，你要让某些东西真正在一家公司里 部署（Deployed） 下去，以此把你的钩子挂在他们身上。因为再说一次，产品本身不是最大的事情，关键是先进门，然后找出什么是有价值的。这就像是你处于“永久测试版（perpetual beta）”或者“永久精益创业”的状态。我认为，在我们向“人类+智能体经济”转型的过程中，倾听人们在为何挣扎将成为一项技能。首先，这只有在你拥有分发渠道时才可能实现，因为只有当有人给你反馈时你才能得到反馈。但因为反馈循环是如此紧密——比如有人给我提了一个功能点子，我可以连夜做完，第二天早上就部署上线。所以是的，我们的思考方式是，我不并想坐在那里凭空发明东西，我想让你告诉我你想要什么？什么能让你的生活更轻松？你现在正在做什么事，如果我帮你解决了会很有价值？所以，不仅仅是发布新东西，而是高频地发布人们要求并且愿意付费的新东西，我认为这是另一项将变得超级有价值的技能。我们将在这上面花很多时间，这应该会为我们创造一个新的任务流水线，让大家都有事可做。

[原文] [Steve]: Then the last thing is data So the the only thing that's going to be ever in this humans plus Asian economy is going to be the data that you collect while people are doing that work So if you can figure out the new ways of working for people like one of the things we're launching with humans plus agents is the ability for people to track here's how we're doing work today and is it is it manual is it done with an SOP is it done and as we collect that data which is useful for the company to know what work is not being done with agents that could be done with agents so that we can create a pipeline to do that but that's data that we will collect and we will know sooner than most other people how companies are doing that work and so we will be collecting just a really valuable store of data as well So right that's kind of the way we think about it I don't know how it all plays out and what we do but I very adamant that we focus on only doing things that people ask for and that add value in which in our world is would you pay more for a product if we built it

[译文] [Steve]: 最后一点是 数据（Data）。在这个“人类+智能体经济”中，唯一长存的将是你收集到的关于人们如何工作的数据。所以如果你能找出人们工作的新方式——比如我们在“Humans + Agents”发布的功能之一，就是让人们能够追踪：我们今天是如何工作的？是手动的吗？是有 SOP（标准作业程序）的吗？当我们收集这些数据时——这对公司很有用，因为他们可以知道哪些还没被智能体完成的工作其实可以用智能体来做，从而创建一个自动化流水线——这也是我们将会收集到的数据。我们将比大多数人更早知道公司是如何进行这些工作的，因此我们也将收集到一个非常有价值的数据宝库。这就是我们的思考方式。我不知道最终结果会怎样，但我非常坚持我们只专注于做人们要求的、并且能增加价值的事情。在我们的世界里，判断标准就是：如果我们做出来，你会愿意为此付更多钱吗？

[原文] [Jonathan]: Yeah And again one thing I admire about you Steve is that you're always aware of making sure that things aren't just fluffy but actually have some coherent impact and value on things That's always been something I've loved and respected about you One qu actually two questions One should be an easy one You talked about the black belt offering Tell me about that for people who want to do with that with you

[译文] [Jonathan]: 是的。再次强调，我钦佩你的一点是，Steve，你总是清醒地确保事情不仅仅是虚浮的，而是真正具有连贯的影响力和价值。这一直是我喜爱和尊敬你的地方。一个问题，实际上是两个问题。一个应该很简单。你谈到了“黑带（Black Belt）”产品的福利，跟那些想和你一起做这个的人说说吧。

[原文] [Steve]: Yeah so we've been doing training over the last few years is how we started the work that we're doing And we found that this one particular training that we do is our AI workflow engineering training that most people don't really get what AI can do for them in their role until they have an aha moment And so we realize that the there are no metal words that we can say to you that will get you to have the aha moment It has to happen with you using AI which is something that you went through either before working with us or after I don't know when you had your magic moment but there there's a moment where you're like "Okay I get it." What we've done is we're making that available to everybody on a trial basis so you can go in and have that magic moment or not Like just find out if it's for you And that's usually the thing that gets people to really dive in and take the next step So yeah that's why making that available to folks

[译文] [Steve]: 好的，过去几年我们一直在做培训，这也是我们开展工作的起点。我们发现我们有一个特别的培训，叫“AI 工作流工程培训”。大多数人直到拥有那个“顿悟时刻（aha moment）”，才会真正明白 AI 能在他们的岗位上为他们做什么。我们意识到，没有什么华丽的辞藻能让你获得那个顿悟时刻，它必须发生在你实际使用 AI 的过程中。这也是你经历过的，不论是在和我们合作之前还是之后，我不知道你的魔法时刻是什么时候发生的，但总有那么一刻你会觉得：“好吧，我懂了。”我们所做的就是把这个培训以试用的形式开放给所有人，这样你可以进去体验那个魔法时刻——或者没有，看看它是否适合你。这通常是让人真正投入并迈出下一步的关键。所以是的，这就是我们要把它提供给大家的原因。

[原文] [Jonathan]: So grateful that you're willing to let people know Of course they're happy to Is there a link you want me to include or what do I want to make sure I have for the people if they want to do that I guess let's just do this live What do you want the link to be I'll send it over to you tomorrow and by the time I will be just the GTMAI podcast and we'll make that the extension then that way I'll send it over and make sure people get it which would be great Perfect All right then And I can tell you there's been as far as so you know my journey is I've had a lot of people who have been key moments in my AI journey Obviously one was AI guy who built his own model and had another buddy who was was the seller in there who taught me some things about it and then you guys were another one because it was a totally different view from outside the company that was refreshing So I had these little moments of different people three years ago I was like ding it and just sparked something and then I just took off from there So definitely you weren't the first but you were definitely a key pivotal part of the journey which I'm again just very grateful for So it's pretty awesome

[译文] [Jonathan]: 非常感谢你愿意告诉大家。当然大家会很乐意。有什么链接需要我包含进去吗？或者我想确保大家如果想参加该怎么做？我想我们就在直播里定吧，你想把链接设成什么？（Steve: 我明天发给你……）等到节目播出时，我们就把它设成 GTMAI 播客的后缀，那样我会发过去确保大家能收到，这太棒了。完美。那好吧。我可以告诉你，就我的旅程而言，有很多人构成了我 AI 之旅的关键时刻。显然一个是那个建立自己模型的 AI 专家，还有一个做销售的哥们教了我一些东西，然后你们是另一个关键，因为那是一个来自公司外部的完全不同的视角，让人耳目一新。所以在三年前我有过这些来自不同人的小瞬间，就像“叮”的一声，点燃了什么，然后我就起飞了。所以虽然你绝不是第一个，但你绝对是这段旅程中非常关键的枢纽部分，对此我再次表示非常感激。这真棒。

[原文] [Jonathan]: Awesome La last question You said a quote a lot in the school and I'll end with this about Steve Jobs or not Steve Jobs No it was sorry dear Bezos and when someone said what's going to change the next five years and he says it's not not the question to ask The question is asked what's not going to change like I remember you said that many times like I love that question So now that we're here what do you tell people because of all the things that changed and all we get is change What do you feel like is not changing over the next however long you think

[译文] [Jonathan]: 太棒了。最后一个问题。你在学院里经常引用一句话，我就用这个来结尾。是关于 Steve Jobs 的——噢不，不是 Steve Jobs，抱歉，是 Jeff Bezos（贝佐斯）。当有人问未来五年什么会改变时，他说这不是该问的问题。该问的问题是：什么不会改变？ 我记得你说过很多次，我爱那个问题。那么既然我们到了现在这个阶段，你会告诉人们什么？因为万物皆变，我们面对的全是变化。你觉得在接下来不论多久的时间里，什么是不变的？

[原文] [Steve]: Yeah and it's a great I haven't thought about that one in a while It's a great it's a great question Like I I think people and companies like if you're in business and you want to do business with other people that you need to continue to focus on it's like business 101 type of stuff like vitamins versus painkillers like people want their problems to go away and so the understanding of what their problems are and how you can utilize whatever technology is available to you to solve them is how you build a business I know it's very it's very cliche to say it in that manner but that's the thing that will not change and I think that's the one thing like everybody's really worried about if I am not adding value to a company based on the 20 years of my experience like what am I going to do is well you you for 20 years you were adding value to a company you just need to find a new way to add value to a company and it happens to be way easier to find new ways to add value to a company if you just pay attention to that and not the tools That's the one thing that everybody gets so caught up in like gamma and like how do we make better powerpoints Nobody cares about making better powerpoints What they care about is communicating ideas better And so if you can find a way to do those things for companies then you're going to be fine I feel like it's a very unsatisfying answer I think I would probably

[译文] [Steve]: 是的，这……我有一阵子没想那个问题了。这是个很棒的问题。我认为人和公司——如果你在做生意，想和别人做生意——你需要继续关注的，其实是“商业 101”那种基础的东西。就像 “维他命 vs 止痛药” 的理论。人们希望他们的问题消失。所以，理解他们的问题是什么，以及你如何利用任何可用的技术去解决这些问题，这才是你建立业务的方式。我知道这样说非常老生常谈（cliche），但这正是不会改变的事情。这也是每个人真正担心的一点：如果基于我 20 年的经验我不再能为公司增加价值了，我要做什么？好吧，你过去 20 年都在为公司增加价值，你只需要找到一种新的方式去为公司增加价值。如果你只关注这一点（解决问题）而不是工具，你会发现找到增加价值的新方法其实容易得多。这就是每个人都容易陷入误区的地方，比如 Gamma，比如“我们要怎么做出更好的 PowerPoint”。没人在此乎能不能做出更好的 PowerPoint，他们关心的是如何更好地传达想法。 所以如果你能找到一种方法为公司做到这一点，那你就会没事。我觉得这听起来是个很不令人满意的答案，我想我大概……

[原文] [Jonathan]: No I think it's correct or we could do but I think it is but it is like everything's changing but still at the end of the day people will pay money for things that will add value to their business and I usually need to continuously remind myself of this it's like how do you help them make more money and it's not about the tools it's not about the anything other than they probably want to make more money and how do you help them do that yeah how you impact business love it man well Steve thank you so much for today thanks for being a mentor tour my journey of AI just again very grateful for that and then when this comes out we'll make sure the link that you'll make for us which is very good very kind will be in the show notes people can go through there and just want to tell you thank you appreciate you so much

[译文] [Jonathan]: 不，我认为这是正确的。哪怕一切都在变，但归根结底，人们只会为那些能给他们业务增加价值的东西付钱。我通常也需要不断提醒自己这一点：你要怎么帮他们赚更多的钱？这无关工具，无关其他任何东西，只关乎他们可能想赚更多钱，而你如何帮他们做到这一点。是的，你如何影响业务。我喜欢这个观点，伙计。Steve，今天非常感谢你，谢谢你成为我 AI 之旅的导师，再次表示感激。等到节目播出时，我们会确保把你那个非常好、非常慷慨的链接放在节目备注里，大家可以通过那里访问。只想对你说声谢谢，非常感激你。

[原文] [Steve]: yeah thanks for having me on this great to see you again

[译文] [Steve]: 是的，谢谢邀请我上节目，很高兴再次见到你。

[原文] [Jonathan]: yeah man you too thanks for listening to the GTMai podcast please like subscribe and come back for More untamed Jonathan

[译文] [Jonathan]: 是啊伙计，我也是。感谢收听 GTM AI 播客，请点赞、订阅并回来收听更多内容。我是狂野的（untamed）Jonathan。