Skip to content

By playing, you agree to uListen's Terms and Privacy Policy and YouTube's Terms of Service.

The Calum Johnson ShowThe Calum Johnson Show

AI Runs My Entire Business: EASY Way To Use AI Agents In Your Work, Content & Life (Step by Step!)

Rachel Woods is Ex Facebook AI & is the Founder of AI Momentum Protocols. She runs her entire business with AI Agents. Her AI even did eight hours of work in her business while she took a sick day.

Rachel WoodsguestCalum Johnsonhost
Sep 14, 20261h 29mWatch on YouTube ↗

Summaries, chapters, transcripts, and FAQs on this page are made by uListen, not YouTube.

EVERY SPOKEN WORD

  1. 0:00 – 2:02

    Intro

    1. RW

      It is so doable to get AI to do more for you. My AI did over eight hours of work without me having to be the one driving it. Things from, like, managing all my inboxes, rescheduling meetings. It found, like, new business, sent out, like, the right Stripe link to someone, so, like, made money technically for me.

    2. CJ

      But you said, "The biggest problem I see with most people is that they ask how to build their first agent when they should ask how to build their first playbook."

    3. RW

      Think of a playbook as just step-by-step instructions. Whenever an AI follows your playbook, it will do work the way that you want it done. So let's actually walk through what running one of these playbooks looks like. This one playbook alone tripled the number of projects that me and my team do. I am going to start in Claude Chat.

    4. CJ

      Before we get into the episode, I have to say that it's pretty surreal to me that I get to do this and record this show for you guys, and so many of you tune in and leave comments and likes on the content. The reason why I do this show and I record every single week is that I want to show that anything is possible. Any dream, any delusion that you have in your mind right now, you can bring it to reality. One of my goals at the moment is that I have this crazy delusional dream that this time next year we'll reach a million subscribers on this channel. And so if you click that subscribe button, the promise that I'll make to you is that we're gonna improve this show every single week. And so I wanna take you on that journey with me. So hit that subscribe button and let's get into the show. Rachel, if someone's watching this video with this feeling that when they're using AI in their lives, that they're being so much less productive than other people that they see online, they're seeing people working with a ton of agents and these complex, like, AI workflows, and they just feel behind, and there's an anxiety that comes with that. If that human being is listening to this conversation right now, what is gonna be the value if they listen till the end?

  2. 2:02 – 15:05

    Watch this if you feel behind with AI

    1. RW

      Well, I think first you can realize that you're not maybe as behind as you might feel, and then second, that it is so doable to get AI to do more for you. It's just about understanding, like, the mindset shifts oftentimes, and then the structure, the way to think about your work to then get AI to do so much more. And so ... And that sounds abstract, but what I'm really excited about, of what we've prepped today, is going through our exact method of playbooking and, like, how do you actually get AI to go do so much more for you and setting all that up. So at the end of today, everybody who's listening is gonna have really that framework and know how to go do that.

    2. CJ

      Yeah. You know, uh, Rachel, one of the reasons that I was so excited for this conversation is that I think when you spend time online and in this space, there's so many people that are, like, quote, unquote, "experts," but they just got into talking about AI, and you've been doing this for nearly a decade. And so one thing that you even shared with me before we started talking, and you said this, you said, "When I was at Facebook and I first started using this flavor of AI, I was mind blown, and it took me a couple of months even to wrap my head around it." And like I said, you've been in this space for nearly a decade. And so I want you to just rewind before we get into, to playbooks and AI agents and how people can be more productive using them in combination. Can you just go back and give us that context of, like, how you even got started with AI, and then what specifically got you so excited that 10 years later you and I are having this conversation in New York and you still have that same level of excitement? Like, what is it that you saw?

    3. RW

      Yeah. Um, so I guess I like to describe it as I feel like I was just a nerd, and then I became a cool nerd because the world started caring about all this AI stuff, and it also got way more powerful, right, versus whenever I first started in it. Um, but I think helpful context for people is, like, I don't have an engineering background, a computer data science degree. Like, I taught myself everything that I know in AI even. And, um, I ended up landing a job at Facebook in their AI research team. And because it was a research team, we got to work with some more, like, experimental type of, you know, technology. Um, and just how it happened, I ended up working with the flavor of AI that now is the same flavor of AI that powers ChatGPT and Claude and all these tools. Um, they're, you know, language models, right? Everybody's maybe heard of large language models.

    4. CJ

      Yeah.

    5. RW

      These ones, we weren't even calling it that yet. We're like ... You know, in hindsight, they're probably pretty small language models. But, um, yeah, just from that role, I had exposure to what was k- kinda coming up and new in AI, um, but then became so hooked and have just been working with this flavor of AI ever since, so.

    6. CJ

      Yeah. You know what? It's like, um ... And I'm sure you even have, like, a, a deeper perspective on this. It must be surreal for you in a sense because the reality that you just described of, like, when you first started using AI is worlds apart I'm sure from, like, when you're working with clients today or even, like, in your own work how you're using AI. Just to make that so clear and tangible for people, can you kinda just describe those two worlds of, like, you kinda gave the, the essence of what it was back then.

    7. RW

      Yeah.

    8. CJ

      But speak-

    9. RW

      The, the very nerdy side.

    10. CJ

      Yeah. [laughs]

    11. RW

      Right? [laughs] Yeah.

    12. CJ

      But speaking today-

    13. RW

      Yeah

    14. CJ

      ... what is AI ... Like, what are you seeing even when you're working with clients and even in your own work in terms of what AI is actually capable of doing?

    15. RW

      Yeah.

    16. CJ

      And how far is that removed from where it began for you?

    17. RW

      Yeah. So that's where I feel like just the way that my life has worked out, it's been this, like, perfect storm because, yes, there's a super nerdy side of, like, how it works, but, um- The other kind of piece of me is I've been an entrepreneur since I was 15. So my very first business was a photography business, and p- my clients would have to pick me up in their cars and take us to the locations, which my parents loved, I'm sure.

    18. CJ

      [laughs]

    19. RW

      Um, but, uh, yeah. And I, I learned about how to do marketing and, like, all this stuff through that business, and so I've just kinda been this, like, business builder wanting to create things, and then also so curious about technology, and I get a lot of that from, like, my dad and my family. And then that kinda coming together, um, what ended up happening for me was when I was at Facebook, I was very in the weeds on the tech, and I, um, ended up actually really falling in love with the wine industry out in California, 'cause I was living in California at the time. And in particular, the people building businesses in the wine industry. And it ended up just being this aha moment of like, well, wait a second. If AI understands language, and at that time it could start doing things, right? So you could, like, give it a prompt and it could write an email for you or write a marketing copy or do all these different things. And then the other side of my passion was, like, really helping these businesses that had become a lot of my dear, dearest friends. Um, I ended up seeing, like, oh, actually, if we could combine AI and business in this way, maybe there's something there. And this, again, was before ChatGPT. Um, so I actually ended up leaving Facebook, raising millions of dollars to build an AI-enabled winery platform for winery small business owners to help them grow and market their business 'cause they wanna spend their time in the vineyard. They don't wanna spend their time, um, you know, nece- necessarily marketing or writing emails. So we were using AI to do it. Uh, somehow just those two things coming together has made me just obsessed with this very specific, um, problem of like, how do we just get way more done with AI helping us take things off of our plate?

    20. CJ

      Yeah. You know, I actually can't imagine what that experience is of you're super early to this technology, and you almost have this, like, light bulb moment that this is gonna change the way not only, uh, we operate in our day-to-day lives as humans, but specifically for entrepreneurs. 'Cause you even mentioned your background is that, like, that's really how it started is you as an entrepreneur. Like, this is gonna change how we run businesses holistically for everyone.

    21. RW

      Mm-hmm.

    22. CJ

      I can't imagine having that thought almost, like, independently. Like, probably with, like, a small group of people, other people have that same thought.

    23. RW

      Yeah.

    24. CJ

      And then it's like the world catches on. And then all of a sudden it's like, um, I think speaking to you today, so many people are trying to figure out, like, everyone's trying to figure out, how do I use AI in my business to run more productively, to save time, or to make money? And so I really wanted to go back because there was something that you mentioned when we spoke a few weeks ago that I wanted you to elaborate on. And you said, "I help train people on how they can be the ones that are operating with AI. And I show non-technical ambitious people how doable it is." And there was a couple of things there that I wanted to key in on, but the first thing is how they can be the ones operating with AI.

    25. RW

      Mm-hmm.

    26. CJ

      That wording felt very intentional to me. Why do you say it like that, operating with AI? What does that mean?

    27. RW

      Yeah. Well, first of all, like, I think what a, like, pointed observation that you even just have around it has been weird, just being really candid. It's been weird being like, I think that we're going this way with how we work, but also the technology hasn't really been there. Um, and to put that more concretely, like, I think the, the way the tools work today, it is truly doable for anybody. But even when ChatGPT first came out, like, getting to this state of the business runs the way we, you know, say we want it run, like, that wasn't doable. Um, and so I would say really the last, like, three to four-year journey has been about figuring out how to make it as doable as possible with the tools. And really s- six to nine months ago, the tools changed a lot where now it is so doable. So it just feels like now is really such a special moment where it has never been this easy to be this person. But the, that's the tool side. The hard part is the thinking. And I would say that is the thing that I think people who've been messing around with the tools a lot over the, the years, that's the thing that they've really honed in on, is the way to think with AI in this kind of more operational way. Um, that if you're newer to trying to figure this out, that's the number one thing to prioritize, is like, how do I actually really start to, like, operate? So what does op- operate mean? Um, to me it's, like, operations is just the idea that you can organize your work so that it works the same way or works as you expect, doesn't feel chaotic, and that helps you get more done.

    28. CJ

      Mm-hmm.

    29. RW

      Um, and having that, like, mindset shift of kind of, again, like, being really operational and, and thinking about, like, a system that you might wanna set up.

    30. CJ

      Yeah. You know, so I, I know that you're, and you're really the first person that I've heard speak about this, which is why I'm excited that you're here. Um, the concept of, like, the playbooks, uh, your playbooking system. Two things I want you to share with people and the person that's sitting at home. The first one is when you say operate with AI, what does that even look like? And then what happened six to nine months ago? Like, what was the shift-

  3. 15:05 – 18:22

    How my AI did 8 hours of work while I was sick

    1. RW

      Yeah, so to give you a mental image of operating, I'm gonna just tell you, like, a, a real thing that happened to me. I, a couple months ago, got, like, extremely sick and, uh, like, you know, middle of the night, waking up every hour.

    2. CJ

      Yeah.

    3. RW

      You know, and you're just like ... You probably have the feeling, like-

    4. CJ

      I've had that, yeah

    5. RW

      ... and you're like, "Wow, there's no way I'm working tomorrow." And I think a lot of us, oftentimes you, like, have this habit of, like, I'm just gonna push through anyways and, like, still try to get, you know, stuff done. Um, but I was just in a state of, like, I just know I, like, can't, uh, can't work. And so I was like, "Well, okay. Um, I know that I've set up my AI operations, my playbooks to build..." There, it was already at that point doing a lot of work for me, but let me, let me really put it fully to the test. And so what I did was I opened up my laptop, and I typed one word, which was todaying, is the word that I use to start my main daily routine playbook. And because I'd already in the past organized all my work into this playbook structure, and we'll talk through, it's actually really kind of a system of playbooks. Like, I've already organized everything. My AI was able to just go and start doing all of that work the way that I want it done without me having to be there, like, in the chat fighting with it or telling it what to do next or really thinking. I just really, like, again, typed one word, and it went off and did all the stuff. And it did, um, I can go through the list, but I went ... It did over eight hours of work without me having to be the one driving it because I'd already thought through the operations of how to do it. Um, things from, like, managing all my inboxes, rescheduling meetings, figuring out the meetings that didn't need to get rescheduled. Um, it used my proposal writing playbook, uh, which we'll, uh, go through today to fix a proposal for a client and send that off. It found, like, new business, um, uh, sent out, like, the right Stripe link to someone, so, like, made money technically for me.

    6. CJ

      [laughs]

    7. RW

      You know, just, like, um ... And there was a lot more. It ran, like, reporting for some of our webinars, built a new funnel, um, all because I'd already, like, set up the operation, set up the structure of the work. So then I don't have to be the one driving anymore. It just, like, operates on its own.

    8. CJ

      Yeah.

    9. RW

      Yeah.

    10. CJ

      So, so when you describe that experience, which is an incredibly relatable one, right? Of, like, it's the middle of the night, and you can feel the sickness of, like, I'm not gonna be working tomorrow.

    11. RW

      Yeah.

    12. CJ

      And so when you say that the, the, the playbook that you ran essentially meant that AI worked for eight hours and did a lot of the tasks that you would've been doing if you were healthy, I think a lot of people wanna get to that space.

    13. RW

      Yep.

    14. CJ

      And simultaneously, a lot of people when they hear that experience, they think, oh, like, agents.

    15. RW

      Mm-hmm.

    16. CJ

      Like, that must be agents-

    17. RW

      Yeah

    18. CJ

      ... are running.

    19. RW

      Yeah.

    20. CJ

      But then you talk about playbooks, and so can you just clear up and maybe those, that, the distinction for people of, like, AI Versus AI agents. And now they're listening to this, to our conversation. They're like-

    21. RW

      Like playbooks

    22. CJ

      ... okay, playbooks.

    23. RW

      Yeah.

    24. CJ

      Can you kind of just clear all of that up and, like, what is the distinction? Was it even agents that were running that process? Like, how does that all work together?

  4. 18:22 – 29:04

    How to actually get AI agents to do good work

    1. RW

      Yeah. So the short answer is yes, it was agents running all of that work, but the agents weren't just, like, willy-nilly on their own deciding what they wanted to do. They were following what I've thought through as my process and my operations for my work, AKA my playbooks. So playbooking really is like the, it's the technique, so it's definitely, like, our term that we use for how to structure your work in a certain way. Um, and what you get out of a playbook is whenever an AI follows your playbook, it will do work the way that you want it done, and that you can actually trust it s- 'cause it's doing it the way you want it done, so that whenever it comes back to you, you know it's correct. So you don't have to, again, like, fight with it, correct it, give it feedback. Once you have the playbook set up and it's working well, then you know that you can, like, hand off that work reliably. And then on the, like, what is AI versus AI agents, uh, I think of it as, like, AI and agents is like a spectrum, and all agents are, are AIs that can do more for you without needing your approval. Like, they're more autonomous. They have more agency. So you have some agents that are, um, pretty s- simple, and they can go into, like, a tool and figure a little bit out before they need your input. And then you might have seen online, like, people are like, "I've got an agent that's run- running for three weeks," you know? And it's 'cause they have an agent that's set up to be able to make way more decisions and have way more agency on its own.

    2. CJ

      Yeah. You know what? Can you ... And, and I, um, I think, first of all, right? Because it sounds like the way that it's working within your company is you have, you're using AI agents, but then you also have these playbooks, like this technique and this system of instruction that the agents are operating off, which means it can just do way better work. Like, the agents themselves can be way more productive.

    3. RW

      Yeah. And we can actually trust them because they do work correctly, which you need if you're trying to do real work instead of trying to just do, like, cool one-off demos, right?

    4. CJ

      Yeah.

    5. RW

      Yeah.

    6. CJ

      And, and one of the things that you even, um, shared, I'm, I'm sure you won't mind me sharing this, is your business, you have a small team. You have 10 people. If you just had to estimate, like rough estimate, 'cause I wanna make this so real for people, the value ... Because you have, you have 10 people, but then you also have agents working with these playbooks.

    7. RW

      Mm-hmm.

    8. CJ

      What would you estimate is even the value to your business in terms of time that you're saving, in terms of, like, quality of work, even in terms of, like, revenue, of actually having this system of agents that are working with these playbooks and then also humans involved? Like, specifically for that agents plus playbooks, like, what is even the value to you of having that set up?

    9. RW

      I mean, it's immense. Like, I, um, even just as you were talking through that, I was even just like, man, I haven't even slowed down recently to really realize [laughs] just how our business could not exist without playbooks. Um, and so, and to make that more, like, tangible, uh, one example is at the end of last year when we were making our kinda like 2026 plans, and again, this was kind of before, uh, and, um, we can get back to the, like, what changed in the tech question, 'cause it definitely accelerated even our own work. But, um, the, yeah, at the end of last year, I had planned in my, like, headcount that we would grow the team by probably three to five people in a certain department. Um, it was around kind of our, our core, like, content and research and student support area of our business. Um, which for p- context for people, we, again, train people on how to learn how to teach AI how to do their work. Um, and so yeah, there's just a lot that we build out, and we, we expected we would need to have, like, a three to five person growth in, like, headcount. Over the beginning of the year, like, um, the first couple of months, I end up playbooking the core parts and responsibilities that I had had on that headcount, and we actually have hired zero people to do that.

    10. CJ

      Mm.

    11. RW

      Um, and part of that is we were able to then to take that investment and put it in other things, which has been great. I mean, everybody on our team playbooks. Everybody is growing what they're able to do every single day because they're playbooking work and doing more and focusing on what's next. But that was a huge change for a business of our size, of thinking we would need to have five more pe- three to five more people, and just be able to have playbooks do all of it. Um, you know, one of the things, just to give it, make it, again, more tangible, as tangible as possible, is, like, whenever we have students go through our programs, um, we found getting really specific advice on not just, like, how to learn how to playbook, but, like, what are you trying to playbook, and what's your draft of it, and how can we give you feedback and tips based on the hundreds and hundreds and hundreds that we've written. We actually have playbooked the act of giving that feedback so that we're able to give, like, one-on-one extremely detailed coaching, and it, there's still a human involved in that, um, process to make sure it's still really high quality. But we would never be able to have something like that for our students at, like, the way the business is set up, like, without playbooks.

    12. CJ

      Yeah. Hearing you talk about it, what comes across is that on the business side, it's helped you get more from, like, the resources that you have and, like, your business be more productive and efficient, but then also just personally for you, it sounds like it's- Saved you time. It's kind of given you more head space and made you personally more productive. You describe the fact that, um, when you speak with people, especially people that are, they're using AI, but they feel like they're kind of behind when they see how other people are using AI.

    13. RW

      Mm.

    14. CJ

      Part of that, uh, problem and challenge is that they're constantly chasing the ne- the new tool, like the next tool. And you even said this. You said, "I think one thing that's really hard is the way that people talk about AI. It makes you feel like you need to chase the next tool. And I think people are tired of that because they're not usually getting the results that they want."

    15. RW

      Mm-hmm.

    16. CJ

      What's the value of playbooks in that respect?

    17. RW

      We have a philosophy, own the playbook, rent the tech, because if you really, again, think of a playbook as just step-by-step instructions that you're giving to an AI that then the AI will follow, and you know that when it follows that, it's gonna do a good job and you're gonna get value out of the work. So once you make that playbook, you can run that same playbook in ChatGPT, you can run it in Claude, you can run it in Claude Chat, Claude Cowork, Claude Code. Like, any of these tools that have an AI that can follow instructions, you can run your playbook there, which means then you don't have to feel, like, locked into something. Like, I know I talk to a lot of people who are worried that they're, they, they were in ChatGPT, they switched over to Claude, and then they're like, "Well, what happens if I'm, like, building up all my skills in Claude," for example, "and then now something else comes up?" If you're playbooking, you're really taking ownership and thinking through how you want the work done, and you're making these, I'll show you, like, documents, that then when you go to another AI tool, you can just take those documents and put them up there-

    18. CJ

      Mm

    19. RW

      ... in that new tool. You don't have to, like, rebuild things 'cause you've made the piece that matters. You've made the playbooks, the operations.

    20. CJ

      One of the things that I noticed, and it was about six months ago, is that people that were, like, deep in AI and, like, these power users started getting extremely excited. It's funny 'cause I even remember the conversation that we had with, um, Ali Miller, Ali, um, a few months ago on the show, and she was working, like, harder than she ever has before because it was like, with all these new capabilities that AI has, it felt like there was this opportunity right now. And it's interesting because if I was going to contrast the way that she would speak about AI and the excitement that she has with kind of like the more everyday person that I would talk to who was just kind of getting into using AI, and maybe they've used ChatGPT or Claude here and there, but it's definitely not, like, a central fixture of their life. That person, contrastingly, just felt a lot of anxiety-

    21. RW

      Mm-hmm

    22. CJ

      ... um, and felt, like, really behind. And so I'm curious to, to, to hear from you of, like, why that disconnect has even come. Like, what is it that's making people like Ali, and it sounds like even yourself, so excited about AI in the last few months and, like, how it's developing? And then what is the biggest difference between how someone like Ali or someone like yourself is using AI versus the everyday person?

    23. RW

      Mm-hmm. Yeah, I mean, so the specific thing that changed was really skills. So if anybody hasn't used skills yet, um, really what it is is a way to save, um, a set of instructions inside Claude has it, like, front and center, but if you're using ChatGPT or Copilot, they also have places where you can save skills. And what's so great about skills and what really changed a lot is that once you save those instructions there, you can have your AI use them whenever you need it. So for us, we save our playbooks in skills because then whenever we want to use that playbook, the AI already has it loaded up. It can use it and start running it automatically. Before all of that, to get your AI to remember things or to have kind of saved workflows, all this different stuff, you usually had to be in, like, an automation tool. Like, maybe some people have heard of, like, a Zapier or N8N had, like, a really big moment. I don't know if you remember that.

    24. CJ

      Yeah.

    25. RW

      Well, those tools are ... They're great for, um, kind of heavy-duty stuff 'cause they have all this, you know, making sure it doesn't break and, like, notifying you and, um, controls, but they take a long time to set up. So if you think about if you're trying to get AI to do work for you and you're trying to build out your operations,

  5. 29:04 – 35:17

    You can build AI employees with Claude

    1. RW

      like we're talking about, um, you have to spend a lot of time making each individual thing. With skills, it's just one file that you save inside Claude or ChatGPT or any of these tools, and that's just so much faster. I can do, do, do, do, do, build up a whole new department for myself if I just playbook, playbook, playbook, playbook, playbook, and that can be something I can do in an afternoon instead of it being something that took me, like, months.

    2. CJ

      Mm.

    3. RW

      So that small tech change made it so much faster. It also made it so much easier because, um, the, once you save the instructions, the AI will follow it, so really the skill is learning how to think through your work and make good instructions, um, is where playbooking really helps, and that's something that's, uh ... Yeah, like, I can help you learn how to do that so much faster than I could help you, you know, learn all these in-depth automation tools. That's just a little bit of a, um ... Yeah, just a little bit further out in terms of, like, the tech skills required for that. Not not doable, but I'm just saying you c- you can get so much done with skills, and that is, uh, yeah, it really changed the pace a lot.

    4. CJ

      Yeah.

    5. RW

      So.

    6. CJ

      From what you've just described, it's like for the people that are- Using this technology the most effectively today, not in the future, today. With skills, and with playbooks, and with agents, it's actually possible to have AI running your business in a way that it's either helping you make money or saving time or be more productive. And so you know what? I just wanna, I wanna get right into it. And I also, to begin, because I know that you're doing content. You have over 47,000 followers on LinkedIn. Uh, you're doing YouTube as well. Like, you're on all the social platforms. You have your, a business that you run. Um, and then you're also just living your personal life. Can you just share high level to begin with, examples of how that system of agents, skills, and playbooks, how that's working together? Like, specific examples of systems that you have within that, and what's the impact for you?

    7. RW

      Yeah. So to talk inside our business, for example, we have a marketing system which has, um, even in there, like, subsystems for all the different things we do in our marketing, from, um, thinking about and writing all of our newsletter to doing our funnels for any of our, like, webinars, um, that we host or any other campaigns. We have playbooks for reporting, um, because I mean, that's actually one of the impacts of playbooking is then think of all the things that you wish you had time to do that it's kinda crazy for you to spend your human hours doing, so you're like, "Okay, it would take me three hours to go and analyze all the data from a webinar," for example. So that's just in, like, the marketing side. In, um, even my own, like, personal life, my personal operating system, I have everything running off of a daily routine playbook, which I call my todaying playbook. Um, and what that does is, like, all the standard kinda like calendar, email, checks, um, our task manager tool to see, like, what's overdue, what I need to pay attention to. But then that playbook also has access to all of these other playbooks that I can go and, like, delegate work to. So the impact, again, is that one, you know, this system I've set up can go and just do all of this work the way that I've already described of how I want it done. So I'm just sitting there kinda like [laughs] watching it-

    8. CJ

      Yeah

    9. RW

      ... thinking about what else I wanna do next, right?

    10. CJ

      Yeah.

    11. RW

      Um, and so it's, it really just completely changes, yeah, how, how it's been working.

    12. CJ

      Yeah.

    13. RW

      Yeah.

    14. CJ

      I, I think there's so many people that are listening that would l- that just feel so, like, slammed when it comes to their work. Like, they constantly feel like there's, like, this microscope, and, uh, there's never enough time in the day, and they're, like, trying to squeeze more in, and they just do not have head space. That would love to be in the reality that you just described. And so I would like for you to share your screen. [laughs] And then also, um, I'm just curious of how it even gets started, of like, we've kind of spoken about the, the playbooks and the agents and the skills, but showing us what that actually looks like. So let's just, let's get into it.

    15. RW

      Okay. So let's actually walk through what running one of these playbooks looks like.

    16. CJ

      Let's do it.

    17. RW

      Um, and I am going to start in Claude Chat, but remember that you can do this in, like, Claude Cowork, Claude Code, ChatGPT, Copilot, like any of these tools. So, um, in order to run a playbook in a tool, you do need to have it saved already. We'll go through that in a little bit with a different playbook, but I am gon- just pull up one that I already have saved. Um, and how I have them saved is in my skills. So I think, um, we can run my planning projects playbook, which actually, um, we haven't even talked about. This one playbook alone, like, tripled the number of projects that me and my team do because it helps with the project planning process.

    18. CJ

      Okay.

    19. RW

      And then it also, like, does a bunch of the work-

    20. CJ

      Nice

    21. RW

      ... which is super nice. So, um, to run a playbook, I can just say, um, planning project playbook. Uh, let's plan a project, um, for hosting a playbooking and Pinot, uh, happy hour party in Austin, Texas.

    22. CJ

      So essentially just any project or anything that you'd wanna do inside of your company-

    23. RW

      We start this way

    24. CJ

      ... this is where it starts.

    25. RW

      Yeah.

    26. CJ

      Okay.

    27. RW

      Yeah. And so what's happening in the background, and it kinda went a little bit fast, but is it went into its settings, and it pulled up the planning project playbook. So then the AI went and read my whole playbook, and it's starting with the first, um, initial

  6. 35:17 – 48:42

    [LIVE CLAUDE DEMO STARTS]

    1. RW

      questions. So, um, part of what I have the playbook do to start is, uh, ask clarifying questions to help me think through the project a little bit more. Um, so for this one, what's the main context? It wants to be a community event. Let's do medium size budget, and then it'll go through. And so you can see here it's starting to follow the playbook. Um, and this is where we can get into this more, but part of really getting AI to do what you want is really deciding, like, what do you want?

    2. CJ

      Mm-hmm.

    3. RW

      So, um, in this playbook, you see it says like, "Clarify the pitch." What we call our projects, instead of, like, project briefs, we call them pitches, so that's like why it's saying that. So these are the types of things as you think through your work you can get really specific, so again, the AI does, like, work the way that you want it done. Um, but yeah. So start following, uh, the process. Um, so following step one, uh, what does playbooking mean here? What are they gonna do? Let's say, um- Let's pick some of these. If I was actually doing this project, I would be thinking about what am I wanting, um, it to plan

    4. CJ

      You know, I, I, I know that a lot of people talk about, obviously you've spoken about agents. A lot of people will talk about, uh, using AI as like a consultant or like as an advisor.

    5. RW

      Mm.

    6. CJ

      And this part of it seems like that's what it is for you, of like I can imagine that most people, even in their business or even if they're at work, like at their job, it's like you... A project lands on your desk and then you're kind of like sitting there initially by yourself, like how do I even go about this?

    7. RW

      Totally.

    8. CJ

      This is like a really easy first step that you've like playbooked out and it's AI almost working as a consultant advisor, like kind of coaching you through.

    9. RW

      Yeah. I mean, a great way to, uh, think about this is if I were to hire an event planner, they're not just gonna like wing it every time they start planning out events, or a project manager, they're not gonna wing it when they start planning out projects. They have a playbook-

    10. CJ

      They have a process

    11. RW

      ... that they follow.

    12. CJ

      Yes.

    13. RW

      Yeah. And so that's why, um, uh, that also gets into like this playbook is set up to ask me for input. You don't have to do that with all of your playbooks. In fact, most of mine just run a bunch of steps like on its own. But this one I want it to plan and kind of like work with me when it's planning the projects. That's why it's like stopping and asking.

    14. CJ

      Mm.

    15. RW

      So anyways, we can go through and, uh, you know, answer the questions it has. Um, once it confirms, here's the pitch. Does that look right? Uh, it looks good. Then it will move into doing step two, which this is another thing again in the playbook of planning a project. Um, one thing I believe in is that your first idea is not always your best idea, so it's better to come up with multiple ideas and then pick the best one.

    16. CJ

      Mm.

    17. RW

      So we playbooked that into this playbook. It comes up with multiple approaches of how to do this project and then recommends the best one. Um, and so this one is saying, "Okay, here's option A, B, C. My recommendation is option A." And I can say yes. Um, okay, so then it's going to step, uh, three. We can just click through these. And what it's doing now with this step is clarifying the timeline and tasks and drafting a full plan. Um, and this is my demo account. I believe I have my ClickUp connected, but we'll see. Um, let's see. Draft timeline and tasks. That's saying, okay, here's the project, here's the tasks. Um, looks good.

    18. CJ

      The thing that comes across even as I just see it doing the tasks and like even the exact idea that you're going through and everything, typically or historically that probably would've taken like a few hours of you like sitting down probably in the beginning of your day when you're like fresh and like really thinking from first principles and like planning everything out to get to this.

    19. RW

      And at least for me, I always forget stuff.

    20. CJ

      Oh, I'm the same way.

    21. RW

      Right? [laughs]

    22. CJ

      Yeah.

    23. RW

      And so then not till like later... So for example, this is our general, um, project planning playbook, but we have an event planning playbook specifically for like our webinars and other community events we host online, and that playbook is really in depth with all the stuff that we've learned about webinars. And so like, you know, don't forget to make sure we have the reminder email like sequence scheduled. Don't forget to do this, this, ask about this. Um, like you can make these playbooks really, really do a lot of work for you over time just even remembering stuff-

    24. CJ

      Mm

    25. RW

      ... which is huge. So, um, but this is what I wanna get to. So again, we're in this project planning playbook. We did, um, step three was drafting timeline and tasks. And for this playbook what I have it do is go through and, um, tell me which tasks it could do instead. So it's going through and saying, "Okay, like designing the icebreaker activities, AI could draft this."

    26. CJ

      Mm.

    27. RW

      Um, you know, reviewing the plan, you need to do it, et cetera. And when I say that looks good, it builds that all into the final document so that not only do I have like the project planned, but then the AI has all the context it needs so that it can just start going and doing those tasks.

    28. CJ

      Mm.

    29. RW

      Um-

    30. CJ

      So you can kind of immediately start to see where the integration even... and we'll talk about this more later. But with agents, like with actually doing work on your behalf-

  7. 48:42 – 58:45

    How to use AI Agents in your content

    1. RW

      system is that I can just, uh, save raw ideas, and then whenever I want to write content, I can just have it run through a bunch of them, too. Um, so you can, you can set this up in a million ways, just depends on, like, whatever your, uh, workflow and preferences are. So, um, so what it's doing is it's following step one of going through, okay, what's the point? Uh, why does it matter? Um, what's your proof? That's what I have, uh, in my playbook. And then, um, I can say this sounds good. And now it's going through and asking me questions, follow-up questions, to make it more in-depth, which is again, what I have in my playbook, ask follow-up questions. Um, I also, because I'm lazy, have it give me a possible answer, right? Like, when you start thinking about your operations, you can have your AI do anything.

    2. CJ

      Yeah.

    3. RW

      So you might as well have it, like, ask questions and give you the answers, too, right? Um, so you say, "Okay, sounds good," or let's say I didn't agree with those, I would correct it. Um, and then this is something that I learned over time, is that in my content, if I'm just making it for my... like, by hand, I often try to think of, like, how am I currently framing this or thinking about this, and is there a more compelling way for the people that are gonna be reading it? And so again, I have my playbook do that, um, which is coming up with different ways to frame what I'm trying to say so that then I make sure I have, like, a way of communicating it that lands. So for this, like, here are five ways we could reframe whatever, um, I'm trying to say around showing up to meetings, and it's even recommending, um, which one sounds most compelling. So I hit one, and then now it's going through and writing it in my-

    4. CJ

      Mm

    5. RW

      ... uh, tone. So, so then I take this, and because I do still care about my content, um, you know, I'm making sure that I'm reviewing it and editing it where I need to, uh, and making sure it's still representative of my work. But, um, yeah, th- this... every single thing that you do in your day can be turned into these little, small time-saver things. And then whenever you start to really think about, how do I, um, set up bigger and bigger systems like this, that's whenever you have more and more time savings. Like, it, it just compounds, right?

    6. CJ

      I want to get into actually how people can build their first playbook.

    7. RW

      Mm-hmm.

    8. CJ

      And there was something in particular, um, that you shared. We were actually, when we were speaking over email, and I thought it was so good and so instructive in terms of how you think about this, because you said the... and I'm paraphrasing, but you said, "The biggest problem I see with most people when it comes to using AI more productively is that they ask how to build their first agent when they should ask how to build their first playbook." And so to begin, can you just share with people, just make it clear the distinction between the two of, like, agent versus playbook, and then why is that question of how to build your first playbook actually the better question to ask than how to build your first agent?

    9. RW

      Yeah. So if you're using, um, let's say, I'll show you in Claude Chat, right? Whenever you type something in here, um, let's say, like, uh, find me a, you know, wine bar in, uh, Midtown New York City that has reservations open. This is already an agent because it's going off autonomously and doing things for you and finding things, um, using tools, like it's searching on the internet. And so if you're using really any AI tool that goes off and does stuff, you're already using agents. So, like, you're actually already there. Now, I think the next question is, well, uh, are you 100% happy with what you are getting out of the agent? Well, oftentimes, no. We want it to be more reliable, or it's not quite doing work the way we want, or for me to say I have agents run off and do eight-plus hours of work, that sounds crazy, right? The way that you can get to the place where it's doing what you want is by giving it a playbook to follow.

    10. CJ

      Mm.

    11. RW

      And so that's why, uh, yeah, I think a really, really easy step, super doable for everybody, is to go through what we have to go through, taking something in your work, breaking it down into a step-by-step process, and then giving that to agent. And that's it. Then you've playbooked your very first thing, and then that's gonna become the snowball that's gonna help you do tons of other stuff.

    12. CJ

      Yeah. Okay. So, so, so when you say the, the point that you mentioned earlier about you were able to do, like, it was able to do eight hours of work-

    13. RW

      Mm

    14. CJ

      ... for you, but then also, and again, I see people post about this online, and I'm like, is this just a hype thing of they have multiple different agents working together and, like, handing things off to each other. Would you say that's near on impossible unless you have some sort of playbooking system in place?

    15. RW

      It's definitely possible, but you're just not gonna be able to control what it's doing.

    16. CJ

      Mm.

    17. RW

      So, um, I'll give you an example here. I, um, I'm pretty sure Claude Chat lets you do sub-agents, but we'll see. So, um, so okay, great. Now can you spin up sub-agents to go and do this research for downtown wine bars in 10 different US cities? And so-

    18. CJ

      As it's doing that-

    19. RW

      Yeah

    20. CJ

      ... can you explain what is subagents?

    21. RW

      Yeah. So that's when you're saying there's a, an agent that's handing off stuff to other agents-

    22. CJ

      Okay

    23. RW

      ... and they're all working together. Yeah, I'm using the terminal. I also use GitHub. Yeah, so we are in Claude Chat, and, um, you know, I'd asked, "Find me wine bar, um, in midtown New York City." So now I'm in Claude Code in terminal, and if I just ask that same thing, um, of course it's gonna go and use an agent, um, to search on the internet and find those wine bars. And then one of the things that's probably helpful for people, uh, to know is I have it right now, as you can see, like, asking me a bunch of questions and for permission and stuff. Um, you can, in Claude, tell it to not ask you for permission, and that's one way you can get it to go be more autonomous for longer. You can also, in your playbooks, tell it to not stop and ask you for feedback, just keep going, and that's another way to get it to go for longer.

    24. CJ

      Yeah. So it's more autonomous, kind of, in the way it works.

    25. RW

      Mm-hmm. Yeah. Um, so okay. So just found a bunch of, uh, wine bars around here. So now if I say, "Okay, great, now can you spin up subagents to go and do this research for downtown wine bars in 10 different cities?" I can even say, like, "Then when it comes back, um, uh, compile into a, uh, top 10 across the world, um, or across the US list in this agent." Um, what Claude is gonna do is spin up another, like, copy of itself to go work on that for 10 different cities.

    26. CJ

      For 10 different cities.

    27. RW

      Yeah. And so then a question I think people, um, uh, or, like, there's a question of, like, why would I even care about subagents? Is it just 'cause, like, it sounds cool? Well, for this, it's because we're gonna get the work done faster because we're letting 10 AIs go out and do the work all at the same time, um, versus if we had to do it one on, one by one, then that would be slower.

    28. CJ

      Yeah.

    29. RW

      So yeah, so as you can see here, I've asked it to spin up subagents to go and do the research, and so it has 10, now, background agents all going on their own and doing all the research to then come back when they're done. And as I've told it, to compile into a top 10 across, um, the US. So this is... You don't have to, like, build anything. You could go try this right now, and you would already be using agents.

    30. CJ

      Mm-hmm.

  8. 58:45 – 1:03:36

    What should I automate first?

    1. RW

      even if you don't write on LinkedIn, that could be a good first playbook. Because your first playbook is the hardest one. You're learning how to take what you do intuitively and write it in a way that AI can follow it.

    2. CJ

      Mm-hmm.

    3. RW

      So then your second one is so much easier and so much faster. Third, and by the time you're at your fifth, you're gonna feel like a pro, and you're also going to start to see patterns and, like, "Well, wait a second." Like we talked about, you know, a playbook that's gonna research restaurants is really similar to a playbook that could research guests or research potential industries, um, to be in. So I would just, like, start on your first playbook and go, 'cause you're gonna get that momentum. That's gonna help you, um, yeah, r- really actually get your first playbook built.

    4. CJ

      Mm-hmm.

    5. RW

      Um, but okay, if you're really looking for stuff, my favorite four places are any process you currently do. Um, your calendar is a great place to look for playbooks, especially looking historically over the last couple of months and seeing what's recurring in your calendar. Um, think about with a- any meeting that's recurring, what's the prep you do for that meeting? What's the follow-up you do to that meeting? And my favorite thing is which of your meetings could just be a playbook?

    6. CJ

      Mm-hmm.

    7. RW

      There's so many meetings that could just be a playbook. Um-

    8. CJ

      Could you, can you give some examples of, like, even moments where you found that out, where you were like, "We have this recurring meeting-

    9. RW

      Oh, yeah

    10. CJ

      ... and it was, should just be a playbook"?

    11. RW

      I mean, project planning is a, a great high-level example. But recently, the webinar event planning, we used to have a meeting where we would talk through and make all of the decisions live and then make our task list. And we're still using templates, we're really organized and all that stuff, but we realized once we playbooked it, that then we didn't have to be live on a call doing it. So instead, like, now one person on our team runs one playbook, and then they hand off the outputs to the other person that runs the other playbook, and we're done.

    12. CJ

      Mm-hmm.

    13. RW

      Just, like, totally- There's so many things like that. So that's your calendar. Um, your to-do list is another great place. Again, um, repeat things on your to-do list. And then the fourth place is your wish list. Like what would you do if you had unlimited time? Those are oftentimes really great playbooks. Um, but yeah, I just wouldn't overthink your first playbook because you're just gonna get in the momentum and then that's, that's the hard part is your first one, so.

    14. CJ

      Yeah. Okay, so leading directly off of that, 'cause you say ... You make the point that the first playbook that you build is gonna be the most difficult, and I'm just thinking that for anyone that's listening at home, that it's like in that difficulty, they almost like they drop off or they don't build the playbook to the level of efficacy that you have. And so part of the reason why the first one is so difficult, and you even mentioned this, is that we're not used to instructing AI or telling AI how to do things that we do.

    15. RW

      Mm-hmm.

    16. CJ

      And so I know that you have a template of like how you actually build these playbooks and like the components that you go through, and actually moves us really nicely into step two, which is to think through and plan how you do the work. And then you mentioned that there are four components of the playbook. We'll even put it in the description, but can you just share with people what this template is-

    17. RW

      Mm-hmm

    18. CJ

      ... and how they should actually go about filling out this template?

    19. RW

      Yeah. So we call this a playbook outline. So the, the way I think about it is like, don't jump in like you're writing a book and you're just gonna start typing and you're just gonna go and write the whole thing in one go. Think about it like you're gonna outline your work, 'cause that helps you think through it before you get stuck in the weeds. Um, but any type of work that you want AI to do, you just need to come up with these four pieces. First is the trigger, or what starts the playbook. Um, good examples of triggers are either things like a schedule, so every Monday morning I want to run this playbook, or, um, an event, so after every meeting I wanna run this playbook. Those would be examples of triggers, but that's gonna help you figure out what's kinda like the start or of the bookend of, um, the work. Then you have the inputs, which is what are the things I'm gonna give to the AI. Um, and these I like to think of are like you've already trained the person, but what is the stuff you're gonna give them each time they're gonna do the work. So in a, um, meeting recap playbook for example, that would be like the transcript for the meeting, the people at the meeting, the time and date. Those are the things you're gonna need to have as the, um, inputs to the playbook. And then it's really thinking through the steps, right, of like, okay, what do I actually want the AI to do? And like we've gone through today, um, you can get really detailed in like how you want work done.

  9. 1:03:36 – 1:10:54

    These 4 things prevent AI slop

    1. RW

      Like I've, I've tried to point out, um, like I have weird ways that I write LinkedIn posts and that I plan projects and stuff, so that's why I have my steps structured like this. Yours could be just however you want AI to follow that process.

    2. CJ

      Mm-hmm.

    3. RW

      Um, and then your output of what are you actually gonna get out of the playbook, why are you running it? And really like, uh, why I suggest starting there is, again, this is the hard part of actually taking your work and thinking about it in this way, but once you can write down your outline, like everything else becomes so much easier because you know what you want the AI to do.

    4. CJ

      Mm-hmm.

    5. RW

      Um, so yeah. So that's the second step. And then the third step is to go in and actually start writing it, 'cause if you notice in my playbooks, like there's a lot more in here than just the inputs and the steps. I also have all this extra stuff.

    6. CJ

      Mm-hmm.

    7. RW

      Um, we probably don't have time today to go into all of the weeds, but, um, this is a template that I can leave with you guys so that, uh, people can, um, at least follow some of the structure of just seeing how to write really clear steps.

    8. CJ

      Yeah.

    9. RW

      The other thing I like to warn people, whenever you look at these you're probably like, "This is a lot." Like, am I really sitting here and like writing out by hand?

    10. CJ

      Yeah, I was gonna ask that.

    11. RW

      Yeah.

    12. CJ

      Yeah.

    13. RW

      Um, of course you can use AI to write these, right? Um, I use AI to write mine all the time. I recommend your very first playbook, though, actually doing it fully by hand.

    14. CJ

      Yeah.

    15. RW

      Because one of those things that once you have the like tactile feeling and understanding of, then you can review the AI's work and actually know, you know, how to fix it and what you're doing. Um, but you wanna start out, uh, really just doing it by hand so you know intuitively how to do it.

    16. CJ

      Yeah.

    17. RW

      Um, but-

    18. CJ

      That was actually my question. I was like, do you have a playbook on writing playbooks?

    19. RW

      Yeah.

    20. CJ

      Uh. [laughs]

    21. RW

      We have many.

    22. CJ

      Yeah.

    23. RW

      And what's also interesting is like the way ... And now we're getting into a little bit more of like kinda the where this goes, but the way that my, uh, you know, playbook writing playbook works is actually different than like Hannah's on my team. Like, like her, the way that she really thinks through working, um, and breaking down her work is gonna be d- different. It's, it's kind of a personal thing, but that's really when you get into like operating with AI is you start to learn how do you want your operation structured and then build it that way.

    24. CJ

      Yeah. You know, from those four components, so you have the trigger, you have the input, um, you have the steps, and then like the finished output.

    25. RW

      Mm-hmm.

    26. CJ

      Is there one that when you're working with clients you find that they typically either just completely overlook or they don't do it in any way near enough depth? I'm just curious for the person that's doing this for the first time, like what's the, what's the miss or what's the part that they would overlook or just not spend enough time on-

    27. RW

      Mm

    28. CJ

      ... that could actually lead them to getting like a pretty poor playbook?

    29. RW

      Yeah. So these are definitely more of an ... Art than a science, um, I will say. So, uh, I don't want this to discourage anybody from starting to make these, 'cause, like, you're gonna write it, and you're gonna run it, and then you're gonna improve it, and that's just part of the journey. But after you write your first one, some of the things I would go back through and look for, especially if it's not working the way you expect, is people oftentimes have their steps be too big. Um, an easy tell is if you have the word and in your step, do this and this and this and this, or even a list of things. If it's not working like you want, then break it up into smaller steps, 'cause then the AI, um, it's kind of like grapevines, where it's gonna spend all of its energy trying to do the step one step at a time. And so if you have a lot of things you're asking it to do, it's gonna be worse at that. So I would split your steps up. Um, and then the other thing is the more unique to you the playbook is or what you're asking AI to do, the more you're gonna have to explain. So if you have, like, a special process or special acronyms or something, um, those are the areas to spend more time giving it more context or more examples 'cause that's gonna be just unique to you versus other things that are more generic, you know, it's gonna be better at without, um, needing a lot of explanation.

    30. CJ

      On the, on the... 'Cause you even mention, uh, like, examples, 'cause here you have, like, the, the Google document of your, uh, playbook, and it's written out and all the steps. Will you have, like, screenshots and images and, like, how deep does the examples go in terms of, especially when you say making it unique to you and wanting to provide more depth? There's screenshots, there's images. Does that come into it?

  10. 1:10:54 – 1:19:55

    Claude skills is a goldmine!

    1. RW

      and then my output is the LinkedIn post. This is what I suggest doing by hand, is that thinking, that level. So it can be really easy. Then going from this to this, and again, this is because we have a template for people learning how to do this, it could also, like, this is a playbook as well, right? Just, uh, written without it being in a template. You can use AI to help you get there to really just, like-

    2. CJ

      To flesh it out

    3. RW

      ... flesh it out, yeah. Um, but I do all of that in a document because then you're not, uh, making yourself stuck in one of these AI tools. Once you have one of these documents, it becomes really easy to just go into File, Download. You can do any of these, you know, PDF, and grab it. And then there's, um, kind of, like, a way to save this in every tool. In Claude, um, and I can also include this, so you can put it in the show notes too-

    4. CJ

      Yeah

    5. RW

      ... of, like, how to configure. Um, but in Claude, you just, you log in and go into a chat, and we actually have a little, like, prompt-

    6. CJ

      For converting skills

    7. RW

      ... for converting a playbook into a skill.

    8. CJ

      Yeah.

    9. RW

      The reason we use the word, like, configuring is because you could even technically just go into a skill, and you could copy and paste and type it over. This makes it formatted all nice. That's why we like the prompt.

    10. CJ

      Yeah.

    11. RW

      But you've already done all the thinking and all the hard work, and all of that is yours and lives in your document. And so I could set it up here. I could set it up in ChatGPT, like, again, any of these tools. Um, but what it did was it went, and, uh, it created the skill file, um, saying, "Let me know if you have any edits before I finalize." Um, that sounds good. Whenever you're running this, like, we always recommend- Kind of scrolling through it and reading it just to make sure [chuckles]

    12. CJ

      Yeah

    13. RW

      You know.

    14. CJ

      Making sure it's aligned.

    15. RW

      Um, so either you can tell it in chat, and it might just be only the desktop app that lets you do that.

    16. CJ

      Oh, yeah, yeah.

    17. RW

      Yeah.

    18. CJ

      I think you're right.

    19. RW

      Um, but otherwise, it's really not hard at all. You just go to customize, and you can go to add and upload a skill.

    20. CJ

      And then you just drag in the file that you just created.

    21. RW

      Yeah. So there it is, LinkedIn post writer. If I open it up, it's the playbook. And so what's really nice is, like, again, um, I've done all of the thinking, all of the work already. I don't have to... Um, I'm not relying on, like, this AI tool and how, uh, you know, the feedback I'm giving it, and I'm fighting with it and all that stuff. I just have the process saved. And then to run it, like we showed earlier, if you just go into chat, and you go to new, um, you can just say LinkedIn post writer playbook. Um... [keyboard clicking] Oh, of course, can't type when you're [chuckles] on the spot. That's why we, we whisper into our computers these days, right?

    22. CJ

      Yeah. [chuckles]

    23. RW

      Um, but yeah, what it's doing is it's loading up my skill and then running it.

    24. CJ

      It's following the steps. So okay, so just to, to summarize for people, it's typically you're starting in, like, a, a document, and you're even, especially for the first time, you're even, like, writing out the outline of it. You then actually, at a certain point, you have this template. Once you have clarity on the outline, it's cool, it's good to use AI to actually kind of help flesh out certain details. You're then checking that. You then use that together with the prompt that you have for building a skill. You put that back into Claude, and then it's building the skill, and then you save the skill. So at that point, you're, you've created your playbook, and I think what you've just demonstrated is step five, which is run it. But then there's a second piece.

    25. RW

      Mm.

    26. CJ

      When you say, "And when the output's wrong, fix the playbook, not the message," can you explain what that part means for people?

    27. RW

      Yeah. So part of, again, what we're doing here is we're operationalizing, like, we're creating these processes, these playbooks, that should reflect how we want the work done. So especially when you're writing your first one, if it's making a mistake, then the natural tendency is to say, "Oh, um, you know, uh, you forgot to include, um, like, three questions here." Like, let's say I wanted step one to work a little bit differently. It's natural to come in here and try to give it feedback in the chat, but then, you know, if you open up a new chat later, it may remember that, probably won't. Now it's this thing that you're just like, "I feel like I've corrected this, like, [chuckles] a thousand times."

    28. CJ

      Yeah.

    29. RW

      So what's better is to actually go back into your playbook and let's say the feedback that I wanted to give it was like, oh, I actually, instead of asking the follow-up questions in step two, I wanna extract, um, the one point, you know, with follow-up questions, and I wanted the response to say the point and the proof source and follow-up questions, you know, or whatever the change is, I would just change it here. Um-

    30. CJ

      So rather than s- t- rather than correct it in the specific chat, you're actually going back to the original playbook, fixing it there, so that every time that you run the playbook, uh, following on from that, it's correct.

  11. 1:19:55 – 1:28:20

    How to build a team of AI employees

    1. RW

      out which of your subordinate playbooks needs to handle it, whether it's like the payments playbook or the customer, uh, you know, success issue playbook or whatever. Um, and you can keep getting more and more complex from there. And so, um, that's really like how whenever we're talking about that we have playbooks running our business, we build like from the ground up with those. We build all the individual smaller playbooks, and then we build like the manager playbooks as those subordinate playbooks are, um, good and work, and then we kinda keep, uh, building up from there. And that's what my todaying, like my daily routine one is, is it's a manager playbook that every single morning has like a set, set of things that it's looking at, and then based on, um, uh, there's certain playbooks it always runs, like triaging my email, pulling my calendar digest. There's also playbooks that it chooses to run based on whatever needs to get done, um, like executing sales tasks or writing proposals and all that stuff. Um, and that, the way that I have that playbook running right now, it almost feels like I have a chief of staff, right? That again, the chief of staff isn't like winging it every single day. They have a process that they follow of how they run my daily routine, and, um, that's how that playbook is structured. So uh, I did create like a, um, graphic to kinda show like at a high level how I have my kinda like personal OS structure 'cause I thought that that would be helpful for people. Um, and then, uh, we can try running my todaying thing. I was straight up like one of the hardest things I'll say about demoing, especially when you have a business that has r- like real work, is like we have, we're under so many NDAs of all of the things that we can't even-

    2. CJ

      Of what pr- what you can share.

    3. RW

      Yeah. So even like my calendar tomorrow I know has NDAs- [laughs]

    4. CJ

      Yeah

    5. RW

      ... stuff on it. Um, but, um, but yeah, it's, it's so, um... Again, the, the structure for even like a manager playbook is just the same thing of the trigger, the inputs, the steps, the outputs. So if you learn that one skill of playbooking, you can really build out the teams.

    6. CJ

      It carries over.

    7. RW

      Mm-hmm.

    8. CJ

      You know what, you know what really comes across? Because it's like, um, and you let me know if I getting this correct. But it's almost like there's like two different categories of playbooks that you have. You have the playbooks that are very executional, which is like it's a single task, and it's like, "Here's how you do the task." It's the full instructions, and that's what we spent the majority of this conversation speaking about, is these like very executional playbooks. On the other hand, you then have the manager playbooks, which is almost like overseeing all of these individual task, all of these executional playbooks, and then defining the handoffs and like... And so when you describe your morning routine, it makes sense what you say about like you have to start from the ground up, which is you probably have a playbook for how, um, the agent should triage and check your email.

    9. RW

      Mm-hmm.

    10. CJ

      And then as an example, you d- might have another playbook for how it looks on your Slack in the morning and like checks certain channels and puts that all in one place. And there could be another playbook for like how it analyzes... One thing that I always do, I love doing like voice dumps, like a, I'll just speak into my phone.

    11. RW

      Like brain dumps in the morning.

    12. CJ

      Yeah.

    13. RW

      Yeah.

    14. CJ

      Um, you could have another playbook for how it does that analysis, but the manager playbook would kinda be bringing all of these different things together and like running them in the right way so that it gives you then one output.

    15. RW

      So that you're working with the manager instead of needing to work with all the individual playbooks.

    16. CJ

      Mm.

    17. RW

      Yeah. Yeah.

    18. CJ

      Yeah.

    19. RW

      Um, yeah, and actually even just like something that could be helpful for people as you're like thinking about this is, um, you can also have like multi-level management, right? So you could have like, um, like this is an example that we use for planning out, um, some of our like client deliverables. We have a strategist playbook that goes and decides all of the tasks and adds those to our task tool, and then we have our triager style playbook that every day is looking at the task tool and saying, "Is there anything due today? If so, let me, uh, assign it to the playbook that's gonna run it." And then whenever it's done, it, uh, goes and comments back on the task tool, changes the status-

    20. CJ

      Mm

    21. RW

      ... to complete, and then it's done.

    22. CJ

      Yeah.

    23. RW

      Um, so it... And all of that, I know that sounds like a lot, but it really all just starts from bottom up, like building out the individual playbooks that can do the work, and then you at some point are like, "Huh-

    24. CJ

      Mm

    25. RW

      ... I can playbook the fact that I'm getting the AI do the work," and then you keep going up and up and up.

    26. CJ

      Yeah.

    27. RW

      And it's, it's like the-

    28. CJ

      It's like you're, you're constantly replacing yourself almost in the process.

    29. RW

      Yes, which is how people build businesses.

    30. CJ

      Companies, yeah.

  12. 1:28:20 – 1:29:22

    Easiest way to get started with AI agents

    1. RW

      that if there's, like, one person that then listens to this and takes action, like, that, that to me is, uh, exactly what it takes, so.

    2. CJ

      Yeah. Well, first of all, Rachel, you've been amazing. And it's funny actually, something that I said to, uh, Ally at the end of our recording, one of the things that I just enjoy about doing this show and these conversations, getting to speak with people like yourself, is just the level of, like, passion and energy that you have for it. And I think that that starts with, like, the people that you're speaking with, the people that you're working with, even people that consume your content, the effect that it actually has on people's lives in terms of getting back time and being more productive. So yeah, just a big thank you, Rachel, for coming here and sharing everything that you have.

    3. RW

      Well, thank you. This was really fun. And I'm, uh, I'm really excited.

    4. CJ

      So if you enjoyed this conversation and you wanna hear even more stories like this, then just click here. And also my team is gonna put some more videos that you can watch here. Thank you.

Episode duration: 1:29:22

In the uListen app, the transcript lights up word by word as you listen. Get uListen

Transcript of episode Jvt3FxRKDzU

Get more out of YouTube videos.

Read along on your phone, search it beside the video, or ask your favorite AI app what was said.

iOSAndroidClaudeChatGPTChrome