So, you've built yourself an AI chief of staff. The AI chief of staff is actually something that I have been trying to build since I started vibe coding. There's a lot of repetitive things that I have to do. For example, like prepping for podcasts and speaking engagements and stuff like that. And so, I've built out an agent that preps me. Let's go into a specific workflow of how this chief of staff actually helps you with do something. And I think you did a little prep for this podcast. I basically prompted and said, "Hey, let's generate a how AI flow." And so I was trying to get it to tell me what should I actually demo that is going to be part of this chief of staff app. And then what actually ends up getting output is this. It gives me an executive summary. Here are three different ways you could tell me about yourself and web flow. Just the ability to be prepped even in a few minutes ahead of these things can really make your life better. Welcome back to How I AI. I'm Claire Vo, product leader and AI obsessive here on a mission to help you build better with these new tools. Today's episode is all about the AI native executive. We have Rachel Woolen, CPO at Webflow, who's going to show us how she uses AI, day using AI as an executive. She's also going to show us her formula for teaching teams how to use AI, drive adoption, and get them in those new tools with excitement. Let's get to it. This episode is brought to you by Graphite, the AI powered code review platform helping engineering teams ship higher quality software faster. As developers adopt AI tools, code generation is accelerating. But code review hasn't caught up. PRs are getting larger, noisier, and teams are spending more time blocked on review than building. Graphite fixes this. Graphite brings all your code review essentials into one streamlined workflow. Stacked diffs, a cleaner, more intuitive PR page, AI powered reviews, and an automated merge queue. All designed to help you move through review cycles faster. Thousands of developers rely on Graphite to move through review faster so you can focus on building, not waiting. Check it out at graphite.dev.link/howiai to get started. That's graphite.link/howiaai. Rachel, welcome to how I AI. The reason why we are doing this podcast together is we were sitting at a very recent San Francisco based AI event and we were talking about how so much of the discourse is around the AI native PM the AI native engineer basically like the AI native IC and we spent a lot of time on that topic here at how I AI but you and I have been chief product officers and we want to talk about the AI native executive and I think your workflow and the tools that you have built yourself are such a great example of somebody with, as I say, a very fancy se-level title who is actually leaning into not only getting hands-on with these tools, but building things that help them do their job better with AI. So, how did you get here? Because I will tell you, not every executive is as into this as maybe you and I are. So, what kind of clicked in your mind to get you thinking like my job's going to change. I really got to figure this stuff out. I started coding when I was 16 and it's obviously been a lifelong journey of building and tinkering and making mistakes. And so, I feel like I built that resiliency pretty early in life. And now, in my current role, I probably hadn't coded in about six or seven years. And then when all of the new appgen tools came out, I think lovable is like one year old now, I was like, "Hey, I'm going to try this out." And it felt magical. And I'd played around with some of the earlier versions of GPT (Generative Pre-trained Transformer: 一种基于Transformer架构的预训练语言模型). But when vibe coding started coming on the scene, I actually built an app to capture my kids memories and that was the first vibe coded app that I built and I built it in a weekend and it just felt like something completely different than anything I had ever built before. And here we are a year later and this is now I'm sitting in Claude (Claude: Anthropic公司开发的大型语言模型) code all day and this is an app that I'm continually maintaining. You can see I'm running it on local host and I've probably have built dozens of apps this year. We just launched our AppGen (AppGen: Webflow新推出的应用生成产品) product. So I feel like a lot of times when you're building AI原生产品 (AI-Native Products: 深度融合AI能力，以AI为核心设计的产品) you also have to be just in the weeds a lot and making mistakes. And I'll show you like this product is not perfect. This is a product for one person right now. And so that's really how I got started. And I think it's something that now is just part of my day-to-day. This is my daily driver. This is a story I hear a lot. It mirrors my own story. This is probably why you and I get along so well, which is, started coding very early, a lot of self-taught stuff, and then get into these jobs where coding hands-on keyboard wasn't my job. And then in 2023, 2024, it just became easier to go zero to one on stuff. It wasn't that I didn't have technical skills. It was just I didn't have time to do things. And so the ability to do AI assisted engineering or vibe coding was it just brought me back to the love of coding and building things. And so I hear that a lot. I certainly feel that a lot. The second thing I hear a lot that I tell people who are trying to motivate themselves to learn some AI skills is find something personal to work on. You said like this memory app for your kids and do it in a weekend. You will get hooked. I think these vibe coding platforms are some of the most addictive games on the market right now. And so again, you don't have to start with something at work. You can start with something that's personal. And I think once you get hands-on and realize what's possible, it's hard to think that your day-to-day job is going to be the same. So, let's get into what you built because I feel like I need this. So, you've built yourself an AI chief of staff. Tell us more and how you're using it in your everyday life.

Yeah, so the AI chief of staff is actually something that I have been trying to build since I started vibe coding. This is actually the first app that I tried and then it didn't work the first time that I tried it and then I've just been continually trying to build it. And what you'll see here is kind of rotating agents. I've just been ticking away different use cases. And then a big part of this is also, I really want to help my team level up. And so it would be inauthentic for me to tell them, hey, you need to prototype with AI if this isn't something that I'm doing every single day. I completely agree. And one of the things I want to call out about how you're talking about building this software is one, you're building it for an N of one. You're building it for yourself, so it can be really hyper customized. Two, you're building sort of a multimodal interface to this in which it can be a web app or it could be something just like a little bit more close to the metal in terms of in the terminal or in cursor. And then I love this idea of being able to build ephemeral widgets (Ephemeral Widgets: 临时性、一次性使用的软件组件或应用) that are useful that you can just toss away. Like you may never look at this Q4 road map after Q4 again and you may want to do it completely differently the next time. You can build these little personal ephemeral apps and then toss them away when they're not useful to you. And so I think software has become as accessible as documents have if you can lean into some of these tools. And so let's go into a specific workflow of how this chief of staff actually helps you with do something. I think you did a little prep for this podcast. I did. So I will actually run different terminals that are all running Claude (Claude: Anthropic公司开发的大型语言模型). I do this partially because, I want to make sure that I'm not using too much context and then I find that when I'm asking questions that are of a similar ilk, I get better answers back in that same terminal window. But I basically prompted and said, "Hey, let's generate a how AI flow." And so I was trying to get it to tell me what should I actually demo that is going to be part of this chief of staff app, right? So it knows this structure and then what actually ends up getting output is this. So this is it gives me like an executive summary. You know if we were to vamp it would be like hey here are three different ways you could tell me about yourself and web flow. I just love that you can do this on an ad hoc basis get a structured output feel very prepared. I don't know if you're like me and don't tell my teams that work for me. I am a just in time executive which means like the 2 seconds before I'm walking into a meeting that's the time I have to prep for that meeting with very rare exceptions because you are just back to back to back to backto back things. I say sometimes to people the number one skill you need to learn as an executive is improv because you are just dropped into a wide variety of context. You need to be able to think on your toes. And so just the abil I had a chief of staff and just the ability to be prepped even in a few minutes ahead of these things can really make your life better. So let's go through Okay, this is a very specific example. It's sort of our meta example. Yeah. On how we're going to go through the show, but let's talk about your morning triage and how you start your day every day using this AI chief of staff. That sounds good. So in this case, what I actually gave it was a question. I said, "Hey, how can I make my week, my last week better? What could I have done?" Because I think that I don't have a chief of staff anymore. I used to have a chief of staff at another company. And so I knew that this was something that they would be like kind of combing through my calendar and like, "Hey, you're not blocking your energy, right? And you're not like kind of taking care of yourself." And so, the other the main thing that they called me out on, which I thought was really funny. So we just had our customer conference about a month and a half ago, and I haven't taken quite as many customer calls. And it's basically calling me out and saying, you know what, you're not spending enough time with customers, and that's where you get energy from. And so, it's basically doing a time analysis on my external calls. I'm like, how is that possible? And again, it did pick up. It wasn't 100% correct. And so I think that but it was kind of like a signal to me. I'm like, "Oh, this is a place where I need to go and reinvest, right?" And it's like, "You're a CPO with no regular visible cause for contact. Fatal flaw." And I'm like, "Oh gosh." It's very dramatic. Well, how does this actually access your calendar? How have you technically built? Good question. So I went and I got a token from Google Cloud (Google Cloud: 谷歌提供的云计算服务平台) and I basically I'm not going to show it, but I basically have it in my .env file and so but I didn't really set that up myself. I mean I knew that this was the way that I was going to need to construct it because I had built other apps before. But let's see if I can find it. So this is my .env file. I'm not going to show it to you because I'll have to burn all the tokens I have in there. But it basically has a bunch of variables and so one of those variables is my Gmail. And then it walks you through how to actually go and set that up. One of those variables is my Google calendar. And then I give it certain authorizations. So in the case with Google calendar, it can only read my calendar. In the case with Gmail, it can only create drafts. It can read, it can archive, and it can create drafts and label. And so a lot of this is thinking about how do I put the right 安全护栏 (Guardrails: 限制AI行为的保护机制) in place. It's kind of like thinking about software, right? How do I put the right guardrails in place so that I don't make mistakes or the agent doesn't make mistakes? It's kind of janky software, right? Because I'm like, "Oh, this is pretty powerful." And then a lot of times I'll go and kind of take some of the power away when I realize I've given it too much authority to act on my behalf. And so I think that's part of it is also like me populating my own intuition about what feels good in working with agents and what feels like it's kind of like overstepping. Yep. And so, are you coming to this every morning? And this is sort of a meta-analysis of your calendar, which again, I've had a chief of staff we did every Friday where we're like, what are you doing, girl? Like, let's fix this. But tell me how you would use this on a daily basis. So, tell me about my day tomorrow. What can I delegate? And so, this will run for a couple of minutes. And when it comes back, it'll tell me this is what's coming up tomorrow. This is what I think you can actually you can do something else. This is I think the way that I would like to make this smarter is this is who you should delegate to and then I'm still kind of working on my Slack agent so I I can't show you but what I want to do is basically like send notes to my admin and be like hey could we go do this and is this fetching the calendar events through an MCP? Is it some custom code that you wrote? This is fetching it through an API token. I don't think that there's an MCP that is like an official official MCP. But what you can kind of see, this is why sometimes I like looking at my briefing. So, other things that I do here, I actually have like a little note card that I like to make for myself. I'm like, I did prep for you. That's good. And then, I'm basically kind of looking at this. It the thing that I like about it is it makes it just very easy for me to get a snapshot and view it, but I'm still kind of iterating on what do I want this to be like. Let's just take a pause really quickly on your web app while this is running in the background. And what I love about what you've designed is look, we're all running on Google Calendar or whatever, but you want it to look a certain way. You want it to be designed a specific way. You want to have a Can you show us that little note card really quickly? I got this idea from Mark and Dreason about like what are my top priorities for the day and then did I actually accomplish it? And hold your horses real quick. It's also extremely cute. So, for people that are not on YouTube and are listening, it is actually designed like a blue lined note card and it does this little flip. Look at you product person. I mean, this is one of those other things where when people talk about personalized software, they don't get to talk about the little pieces of that you get to put in the app that just make you really happy. Okay. So, sorry. Sidecar on design. So, you have your top priorities. Yeah. I just as a side note, I almost redesigned the app last night because I was like, I wish it kind of felt more like Apple Notes. And I think that's part of what is fun is like, okay, what would that actually be like if I translated this app and I had like a theme for it? And so, there's something really fun about actually getting to design again, but not actually I mean, it takes the amount of time. It's not it's not at the pixel level perfection of the incredible designers on my team, but it does at least convey like there's a certain um Jennaqua of this app that I'm trying to capture to make myself happy. Yeah, I love it. It's so great. Um okay, so here we go. Here we go. Yep. Okay. Your chief staff is very mean. I tell it to be mean to me. That's why [laughter] I want it to keep me in line. I'm one of those people who likes to try to take on too much, I think, sometimes. Okay. So, it's basically telling me to make certain meetings async. That's a great idea. It's telling me to delegate a couple of these. This is also probably a great idea. I don't think I can delegate this one, but good call. It's telling me there's too much clutter on my calendar. This is one of those things where somebody's just like, "Put something in my calendar. I don't know this person." So that's definitely optional and I haven't accepted. And then it's like, can whatever director cover this and send me a summary. Right. And this is really helpful. This is like forcing me to think about, okay, do I actually need to attend all these things that I've signed up for? And can I give myself a break during the day to do some work and maybe get home to my family at a reasonable hour? I love these delegation messages because just as a human, it is very easy to identify meetings that you don't need to go to and then you're like, "Oh god, how do I say I don't want to go to this meeting?" [laughter] So even getting that draft where you could just find I could just copy paste that into Slack is such a useful friction reducer in managing your own time. And again, we're talking about the AI native exec just too busy, over booked, under prepped, really just wanting to spend our time vibe coding if we're being honest. So, I think this is really, really useful. And again, I like this because it's very conversational as a chief of staff might be and really helps you sort of figure out what you can do with your calendar. And then I like how it's mean to you. What's the brutal truth? Let's look truth. Is that [laughter] I'm like, yep. It's because being a senior PM sometimes sounds more fun than I know. I'm like, is that bad? Is that a bad thing? Um, [laughter] so for those again who are not watching, the brutal truth is you're operating as a senior PM, not a CPO. You're reviewing PRDs, approving scripts, and recording marketing videos. I mean, they're calling a spade a spade. This is the behind the curtain stuff. And I tell everybody this about the origin story of chat purity, which is they're like, "Why did you build chat purity?" And I was like, "Man, because no matter how fancy of a title I got, I still was writing PRDs all the time." And so this is really really great. And then I love this idea of like the only thing that will matter in six months are is this conversation. What an amazing way to level up. So, chiefs of staff that are watching here, you need to end every meeting with your exec with the brutal truth. And build this for yourself. Okay. So, this gives you a really good prep for your day. I'm presuming you do a very similar workflow for email as well. What is outstanding? Can you just walk us through maybe just some of the problems that your email chief of staff solves? Yeah, I mean I think that my biggest problem is that I'm I have like 500 emails deep in my email, but it's basically archiving a bunch of stuff that I don't need to it's keeping a few things in the inbox and then it's drafting a few responses and a lot of this is like I've been continually updating this particular agent to know where to draft responses. Sometimes it draft responses, you know, like some random SDR inbounds to me and I'm like, "No, no, no. You don't need to draft responses to this person, but a lot of times it'll be like, you know, somebody from like a partnership type of email will send me an email and it'll tell me, hey, you need to pay attention to this email." So, a lot of it is trying to also know like when somebody's waiting on me and asking for some document to get shared. And so, this is kind of a cursor view. And then I can go through my inbox very very quickly. I think the Slack version of this is going to be an absolute game changer since we operate almost internally in Slack 100%. But that one has been a lot harder to digest even with where the models are at. It's almost too much data to triage. Yeah. So I feel like that's my next that's my next hill to climb. What does that say about what we expect of humans? That we are expected to triage a vast amount of very disperate information with a lot of disconnected context. So disconnected that fancy pants large language models cannot make sense of it. But for some reason, us with our meat brains are expected to be able to do this. And I just think that is such an indictment of how we're expected to navigate the massive amounts of information as people that are working on teams in these asynchronous or semi-synchronous ways. Yeah, it I mean I think it's also become so different during post pandemic and especially if you're working for a remote organization and I think that we're obviously only four years five years into that experiment and so I think that we haven't like the tools haven't quite caught up with what we need to kind of declutter our brains in through that kind of transition. And so I still think we're in the early days of how to effectively do remote work, but that's a different topic for another podcast. The early story of Atlassian is probably very similar to your own. Atlassian knows firsthand the challenges that startups face every day and that the right tools are essential to go from MVP to IPO. That's why Atlassian for startups gives eligible companies up to 50 seats free on the premium edition for products like Jira, Confluence, Loom, Jira product discovery, Compass, and Bitbucket. So your team can use the best-in-class tools to plan, track, and collaborate on work, whatever that work might be. Many of today's most successful startups like Cloudflare, Canva, and Rivian relied on Atlassian for their growth trajectory. And Atlassian wants to give that same opportunity to the next generation of builders. They know how important it is to focus on building the right things early. Whether you're in the sticky note stage or well on your journey, teams at any stage can work smarter together. It's never too early to start with Atlassian. Head to atlassian.com/startups/howi for more details and eligibility. Okay, I want to do one more fun chief of staff example before we go on to our next workflow, which is how do you prep for dinners? Tell me I just this little tab is calling my name. I must know. All right, so I'm going to a dinner tonight. It is called the CXO Collective. And what I did was I took a screenshot of this guest list. And so, I'll show you. It's, I basically literally just took a screenshot of this guest list and then I dropped it in here. So, you can see the image that I dropped in here. And then it read each of them. And it I haven't looked at this yet, so I can't tell you if this is good or not. Buyer beware. It's free. So, and then it went and did search on every single person on that list. And then I have an agent that is basically kind of going and doing more analysis on like this happens to be one of my colleagues who's going to the dinner guy leaf and so you know like he was acquired. So it looks like they're actually identifying the right person. So it what it does is it goes and it does a search on the web then it does a search in LinkedIn then it goes and tries to do like more web search and you know it's like hey this is your colleague that pretty good and it also has like the the other thing that I would say that I have given this is a lot of context about me. So let me see if I can just show you a couple of examples. Yeah. So I have like some general about me documents like this is from like some communication workshops that I went to where this is like kind of how I want to communicate. This is like personal resources about me and they're all I try to do everything in Markdown (Markdown: 一种轻量级标记语言，易于读写和转换为其他格式) just so it's easier for the models to understand it. And then this I actually generated like a I try to generate this maybe once a month of like what are the new Web Flow products off of our release notes so that it kind of knows everything that we've released in the last couple of years. And so then it kind of knows a lot about me and then it's telling me about the venue. And then it's like what are hot discussion topics. And I'm going to like an AI marketer kind of AEO answer engine optimization dinner and then it's like created this full prep doc. So I'm like okay that's great. And then what you can see is I'm like can you add it to my dinner research tab? So the everything that's in dinner research is actually a markdown file. So that's the other way that this has worked really well. I could go and open this up in Cursor (Cursor: 一款AI辅助编程工具) if I could find that file and then but really I just find it easier for something like this to go into my chief of staff and then let's see pretty oh it's going to be a nice dinner and you know it's like these are my priority connections like conversation starters. I'm like this is like kind of the introvert in me or maybe the ambivert trying to be prepped. I love this because I am an introvert. I get invited to these dinners all the time. And this is really great. One of the things I want to call out for how I AI watchers and listeners is I recently did a mini episode on how to generate a Markdown驱动的个人应用 (Markdown-powered Personal App: 以Markdown文件作为数据源和内容展示的个人应用). And so, instead of having to toss these files into a database or into a file store, you can just store them as markdown files inside your repo and then display them on the front end using a markdown renderer and then it looks really nice. So I think again when you're building sort of a personal app for your own research or context just use markdown files as a source of truthful content and display those markdown files in a nice way on the front end. Simplifies things. You don't have to think about a database. And the benefit is then any of those markdown files can be referenced by any of the agents that you use for other things. So, you could go to Cloud Code and say something like, "Hey, grab the dinner prep from that steakhouse dinner I went to last month and remind me of who that person was at Twitter that I was supposed to talk to. I never got to talk to her and I want to send her an email." Like, that kind of stuff can really help just make not only the the software, the personal software better, but also just your use of overall 代理工具 (Agentic Tools: 能够自主执行任务并与环境互动的AI工具) a lot better. Yeah, I feel like it's been a game changer to start using markdown files. And then also start using markdown for even like PRDs. So, a lot of times I'm like updating this app and I have like kind of an ongoing PRD that's like self-documenting this app.

I love it. And so, what do you think the result of you building this chief of staff has been? How has it helped you? Are you saving time? Do you feel like your life is a little bit better? Obviously, you've learned more about AI. Kind of what is the outcome you've gotten here? Yeah, I think that the biggest outcome for me has number one been really being close to the metal in terms of building out AI products and I am the one that is able to have very very detailed conversations like for example I didn't get a chance to this morning but I'm going to go and start using Gemini 3 (Gemini 3: 谷歌开发的多模态AI模型) as part of this app and there's certain types of it's gotten me very close to our code base as well. So for example, I use codeex a lot. Right now I'm only using it on this particular app, but I could go and be like, tell me about this app. How is it built? And I think that that's really important to be able. So I also obviously have access to our rep our monor repo. And so a lot of times like this happens to be for telling me about this repository, right? But this is like a really good starting point for anybody who's going to go build something because then you can start to understand well how is this app that we build actually built and I think that that it was almost something that was inaccessible unless you were sitting there and reading code which sometimes I do but I think that this is very accessible to anybody. This is awesome. And again, you can just add modules, you can spin up new quad code interfaces, and then you can design whatever you want on the front end. It can look however exactly however you want, which I love. Well, I want to flip this to another use case in the second half of our episode, which is a little bit less about how you personally use AI, even though we've we've shown that a little bit. I do want to talk about how you get AI into your team. And so I want to talk about your builder days because I think you know what you're setting a really great example of is I'm a CPO. You say you think of yourself as an IC CPO. I do as well which is like we're going to do the work. We're going to get into it. Part of it is setting an example of how the work gets done. And then part of it is you have no credibility if you don't actually know what you're talking about to convince other people to do something in their job. And so this is a little bit of a different how I AI which is how I AI into an organization and you've done these builder days. So tell us a little bit more about this. Yeah, so we did a builder day last Wednesday. So this is pretty fresh and that's the second builder day that we did in this particular way. So maybe first I'll talk about the results because I think that it's actually helpful to see that. So this is a screenshot from Hex that was a dashboard that shows all of the roles under my team and then after builder day one which was just design we had maybe half the team using Cursor (Cursor: 一款AI辅助编程工具) and what I think is really cool is like very few people were using it before that and then it became this kind of sustained use because we helped people get over that hump and not everybody and then we just did a builder day part two And I'll talk about a little bit about how we actually ran that builder day, but we had design product data science, user research, analytics, engineering, lots of different functions build for builder day. And what we realized in order to get more people over that hump, and by the way, I don't think this is everyone. I think we actually had something like 80 plus prototypes that were actually built that day. Not all in cursor. But, this is like I'm excited to see like does this continue to be sustained? Is this something that people start to use as part of their day-to-day? And I really think about an organization kind of similar to any normal distribution curve where you're going to have, it's the innovator's dilemma where you have the early adopters. And we identified a lot of these early adopters here back in March, let's say. And then those people actually built out the scaffolding for May for our first builder day. And then as we had more and more people working, we actually had like a bigger user group that was inside our organization that were maybe the early majority. And now we're starting to build into the kind of middle part of that distribution or even maybe some of the lagards that are by just trying to reduce the friction and then trying to lean into the people who are really embracing that technology. And I think it needs to be both, a 自上而下 (Top-down Mandate: 高层领导强制推行的指令) mandate. Like I tell my team, hey, you can't get into a meeting with me without a prototype. So there's a mandate, but then it's also a 自下而上 (Bottoms-up: 基层员工自发推动的变革). So I think it kind of has to go both directions. So this is the outcome you're trying to drive, which is like we want to use these tools. We want to feel like not just during this spike, but overall in our day-to-day, those tools are useful to my team. And so explain to us what a builder day is and how it drives you towards that outcome. Yeah, so we're builder day is when our entire organization stopped doing what they're doing and they built a prototype for whatever they wanted it to be. And the goal was really to boost confidence, spark adoption. And so we had a couple of different types of tools. We lit up Cursor, Figma Make (Figma Make: Figma的AI辅助设计功能), and Webflow (Webflow: 一款无需代码即可构建网站的工具). And I'll talk about how we actually enabled the organization around that. And then we actually had different tracks for different types of teams. So we had different assignments for product. We had, you can explore a new product idea. You can validate user interactions, you can test product concepts, you can build a full-end workflow, we so we also had them do a warm-up assignment and then we basically had support channels. So we had a couple of engineers that were on call for this. We ran a judging panel. So like me and the CEO and a couple of other crossunctional leaders were the actual judging panel. We had prizes, recognition, different categories. And then we really focus on like how do we measure success and do this again and so this is something that is evolving for us. This isn't we we definitely aren't like we've we've nailed every single part of this but I was really encouraged to see so many people actually taking one step outside of their comfort zone. Amazing. And so this is something that for the leaders out there that are just looking to like how can I accelerate adoption? How can I up, upgrade our learning and development in the org, not in this ad hoc way. This is 100% the advice I give them. A builder day, a hackathon, and it has to be a combination of tools, access, education, real prototypes, and prizes. People love the prizes. And so if you can make this something really fun, I think it's a super effective way to engage the whole organization and just becoming more AI AI native. And I would I would just ask you, what's the feedback from the team? What's the kind of before and after you're seeing from the team before and after these builder days? Yeah, I mean, so I actually copy and pasted literally the feedback that we got from our survey, so you can go and see it. Most people were not frustrated, which made me really happy, which meant that we had gotten around of the a lot of the technical issues. I think people found it fun, empowering, motivating, and 大开眼界 (Eye-opening: 令人眼界大开，认识到新可能性). I don't think people understood what was possible. Yeah. And that to me was the most heartwarming thing I could possibly imagine. Is all these people I call it getting bluepilled, right? All these people all of a sudden stepped into a new part of their professional journey and that that just made me really happy.

So to recap this for any of the CPOS or executives or aspiring CPOS or executives out there, a couple takeaways from this episode is one, there is just no substitute for getting your hands on to these tools, especially coding things. I think is very important in almost any role. I've been saying over and over and over again, this is the era of the 硬技能 (Hard Skill: 可量化、可教授的技能，如编程). And one of the hard skills I just think is going to become more Table Stakes (Table Stakes: 参与竞争的基本条件或要求) is at least being able to use code to get things done. Even if you can't be an exceptional software engineer, that you can build yourself custom software, especially markdown based that has both sort of 代理访问 (Agentic Access: AI代理能够访问和利用数据或系统资源) properties where cloud code could use it and web UI. You could run it locally if you don't want to deploy anything. It's not a big deal and it also helps you kind of bypass some of the security concerns around using thirdparty tools because you're running things locally, you're using, internally approved API keys. All those sorts of things kind of help you stay 沙盒化 (Sandboxed: 在隔离环境中运行，以限制潜在风险) in your environment. And then organizationally and culturally, these builder days can have a very repeatable structure. And so if people want this, I think you're going to share this repo and this resource with us so you can look at it. There's a repeatable process for onboarding, educating, engaging, and driving sustained adoption of AI in your org through these sort of spikes in builder days. And then you're just having fun. Are you having fun? Because I am having so much fun right now. And that's one of one of the reasons why I'm spending so much time on AI is it's just it's it's better than it used to be for my job. I think it's the most amazing time to ever be doing this job in the entire time that I've been doing it. It's a hard job. I think it's just building products is hard, but it's never been more fun and the possibilities have never been quite so unlimited. I completely agree. Okay, let's do some lightning round questions and then let's get you back to I just have to call it out. Your two cursor windows down at [laughter] the bottom, one of us, as we say. Okay, so first lightning round question. This is hyper hyper hyper personalized software. Is this a product or do you think that just every CPO needs to go out there and build their as I say artisanal farmtotable chief of staff for your app? Where do you fall on this like personal software versus small market SAS for for something like this? I think that parts of this are probably a product. Like I'd imagine that, you know, maybe Anthropic (Anthropic: 一家专注于AI安全研究和产品开发的公司) tries to build this themselves, right? Where it's like everybody's chief of staff, not just for a product leader. That may not kind of get it everything that people want. And so you may end up having kind of extensions off of that or right but I think there's going to be a whole ecosystem around how to actually get work done and what people are doing with MCPs (Multi-modal Conversational Platforms: 多模态对话平台) is just the beginning. It's still like let me tell you like the very early days of being able to access company data. But I do think that this is an app. I just don't know if it's going to be exactly this form factor. Yeah. And then my second question for you, which is a little bit more on the people side, which is you clearly care about AI素养 (AI Fluency: 理解和有效使用AI工具及概念的能力), how are you hiring for that? How is your team reacting to it? How are you evaluating and promoting kind of what's your framework for talent around this? Yeah, I actually tried to share a little bit of that in this app so that you can kind of see that. But I think that AI is almost like you have to look at each of the different dimensions of your team. Whether it's data and insights, whether it's tool fluency, builder culture, 职业发展路径 (Career Ladder: 员工在组织中晋升和发展的路径), interviewing, like you have to look at every single aspect of how you, your 操作系统 (Operating System: 指团队的运作机制、流程和文化) for your team and kind of like grade yourself and be like, how far have we moved down this spectrum? And in our case, like we're in the middle of redoing our career ladder. Because I wanted it to be really clear to people that it's not just fluency, it's like I expect people to lead and own and understand what is possible and push the limits of our product. And it's not just for themselves, it's for our customers. And then I think you have to kind of like reinvent so many different parts of your your practice as a leader. And so that's that's the skill that you kind of need to lead into as a leader is just like 自我革新 (Reinvention: 不断重塑和创新) at all times. Okay. And then my final question for you, I ask everybody. You have dueling claws open too. So this is an interesting question, but when you are prompting and he's just not listening, he's just not giving you what you want or your chief of staff is too mean. This is really funny because usually I see so many AI agents be too nice, but yours is actually really mean. What is your 提示工程 (Prompt Engineering: 设计和优化给AI模型的指令以获得期望输出的技术) technique? Are you an all cap? Your your AI is an all caps AI. So I'm curious if you go you hit hard right back at it. What's your prompting technique? My prompting technique is first when I don't like what it's giving me, I will clear cloud code. So, a lot of times I'll clear it and then I'll try again and if I'm still not getting what I want, I'll be like be a 100x more this thing that I want or, be 10x. And so, it's like I almost use like 10x as like a little bit more and 100x as like go and just rip it apart and do something else. So, this is a repeated tip. We heard this from Hillary at Whoop in one of our early Howi AI episodes, which is these models like numbers. So say be like 20%, nicer or be 100x more harsh or make this three times as purple. Whatever it is, you can quantify it. It can calibrate how it's working for you. So great, great tip. Well, Rachel, thank you for joining us on how aai. How can we find you and how can we be helpful to you? Yeah. Thanks again, Claire. This is amazing. I'm Rachel Woolen on Twitter and you can also find me on LinkedIn to follow me. And I also recommend checking out webflow.com. We have a brand new AppGen (AppGen: Webflow新推出的应用生成产品) product that we just released last week. It's really for your marketing team to be able to do vibe marketing which is a whole new wave that we're riding. So amazing to be here. Awesome. Thanks. And I appreciate you sharing all your tips on how I AI. Thanks so much for watching. If you enjoyed this show, please like and subscribe here on YouTube. Or even better, leave us a comment with your thoughts. You can also find this podcast on Apple Podcasts, Spotify, or your favorite podcast app. Please consider leaving us a rating and review, which will help others find the show. You can see all our episodes and learn more about the show at howiaipod.com. See you next time.