Claude Just Became My Personal AI Coach (Full Setup)

In this episode, I will build a personalized executive coach inside Claude by focusing on planning, orchestration, and a clear architecture. Stay tuned!

**Are you a Coach, Expert, or Service Provider wanting to get more Customers from your Content? Come work with us and get access to our proprietary AI Content Marketing Tool (SPARC) for FREE for the next few days! Text "COACH" to (866) 498-2080 or DM us on Instagram @thebusinesslounge for a free strategy chat!

Timestamps:

00:00 Building an AI Coach

00:27 Business Context Interview

01:53 Personal History Layer

03:08 Genetic SNP Deep Dive

04:51 Not a Replacement

06:34 Project Instructions Setup

07:56 Memory and Dialogue

09:09 Saving Key Takeaways

10:55 Why Claude Works

12:02 Claude Walkthrough

13:59 Files and Protocols

16:24 Wrap Up and Next Steps

📢 Subscribe to the CMH Show on your favorite podcast platform for more insights on AI, advertising, and business growth!
💬 Leave a review and let us know your thoughts or topics you'd like us to cover next!
🌐 Follow us for updates and exclusive content!

#Claude #DigitalMarketing #Advertising #AI #BusinessStrategy

Transcript

SPEAKER_00 0:00

Today I'm going to show you guys how I built a high functioning executive coach using Claude. Let's get into it. All right, guys. So let's check this out here. I want to give you the architecture first because I think it's really important to make sure you know the proper organization and uh how you need to plan this thing. I've learned with AI the best thing you can do is the planning and the orchestration and then the architecture. So really think about what you're doing first. And I mapped this out, and then I'm going to show you guys what it actually looks like on the cloud side. So the first uh thing that I did is I went to a thread inside of a project. I'll show you that here in a second. But what I wanted it to understand was the specific business details, meaning the business that I'm running or the businesses that I'm involved in, as well as some of that historical context. So I need it to at least know what I do in the world, and I need it to know uh the ways that we make money, the people that we serve, so on and so forth. So literally what I typically do in this situation is I flip it on Claude. I start out and say, This is what I'm trying to do. I want you to be my executive coach, uh, and I'm gonna give you multiple layers of information about me, but I need you to first understand what type of business I run or what type of businesses I'm involved in, and then ask it to interview you to get as much information as it needs. Now, it's up to you as a pilot to determine where you're gonna cut that line. So, in other words, it may ask you one question or two questions or a series of questions, and then you it says you're ready to move on and you say yes. So use your own discernment. If you feel like you've given it enough that it has enough context about you, you're the pilot, you'd make be the one that decides. Or say, no, I don't feel like you have enough about me and my business yet. Continue to ask me more questions or ask me more questions about this. So that really comes up to you and also your preference in terms of how much information you really want to feed it. And there's some privacy concerns for some people. Totally get that. I have those as well. And so let that be kind of your decision, right? And use your own discernment there. So that's where I start. Obviously, if it's an executive coach, I want it to have information about my business. Next is historical context. Now, when I say this, what I mean is historical context about me. So, what I really want it to understand are my strengths, my weaknesses, uh, and historical context from the standpoint of what does my upbringing look like? Now, again, personal preference about how much you want to reveal. I typically operate in some generalities. I don't give really specific details about that, but enough to where it has what it needs and it's equipped to serve me in the ways that I need it to serve me, right? So, what are some lived experiences that I've had in my life? What are things that I've been involved in? Was I in team sports? Like, have I had challenges with X, Y, or Z in my life? Have I excelled in certain areas of my life or failed at others? I wanted to know where I'm really strong and where I have gaps or weaknesses. And then what I did with mine, and you're welcome to do this as well, is to have it do some research on other potential entrepreneurs, if you're in that space, that are similar to what you've described. And I found that to be dangerously accurate. Like it found, I think, five or six people based off of the deep work we did together with the historical context and the business layer that it pulled and it allowed me to choose which one sounded the most like me. And then we built an entire operating protocol, personal operating protocol around that, which was super, super helpful. And I'll show you that in a second. And then the third layer is your genetic SNP. Now, you don't have to do this. This is like next level nerd. Uh, this is my Stratagene. So if you're not familiar with what Stratagene is, they run a genotype report. So if you do something like Ancestry or 23andMe, you actually give it your raw data to Stratagene, and then Stratagene will give you all of your genetic SNPs, which means genetic mutations. Now, what that means is you might have a propensity where for me, I burn dopamine faster than most people. I have what's called a fast COM T genetic mutation. So for me, I can really build a lot of addictive tendencies, which is either good or bad. Uh, but for me, I'm always seeking new dopamine, and I have to build how I operate in my day, both personally and professionally, around my need for more and more dopamine. So thrill seeking, um, it's hard for me to do monotonous tasks. I get really bored with that, more of your ADD type personality. Also, I have uh methylation restrictions due to my MTHFR gene. I think like 53% of European Americans have this, so it's very likely if you're watching this and you fit that description that you're also in that category as well. Methylation is really important. It's basically the process by which you take what you consume and convert it into things your body can use. That's like the easiest way to explain it. And when you have that, unfortunately, there's certain foods like folic acid, if you consume that, it's actually going to throttle down your methylation even further. So I do things to encourage, they're called methyl donors, to encourage better methylation because I know I have that restriction. So I fed it this data because I wanted that to be factored into both the historical gaps in the historical context, strengths, weaknesses, as well as business context, because now I can operate dynamically. So the way I outlined it here for you is it's like I have a therapist, a business coach, and a functional medicine wrapped up, funk med professional wrapped up in one. Now, caveat to this, and I want to be very, very clear when I say this this is not a replacement. This is uh an add-on, right? So I have all these things. I have a functional medicine practitioner that I work with, I have a business coach that I see once a week, and I have a therapist that we go to on a regular basis as well. So what I'm suggesting you use this as, and really important, I cannot state this clearly enough, is that you're using this on the in-between stuff. So if I see my business coach every Thursday and I have profound takeaways and insights, I add that to what I'm doing here, or I'll add it to it with some restrictions, right? Some privacy restrictions as far as what I feed it. But what I want is I'm cutting the gap, right? So I'm not having to wait another week. I can continue dialogue and then I bring stuff to my business coach back the next week based off the homework, right, that I did on an in-between basis. I recommend the exact same thing with functional medicine with if you decide to do something like your genetic SNPs or even just anecdotally, right? These are observations you've made when you eat certain foods or behave in a certain way or take certain supplements, you anecdotally notice a benefit or uh or it's inhibiting. So that's totally up to you. I wanted to go the full nerd route and I gave it my strategy and report. You don't have to do that. But if you are going to go that route in any of these, I recommend arriving at confluence, which means that you're basically going to ask another AI model to audit the findings of like Claude, for example. So for me, I gave the genetic SNPs to Claude, and then from Claude, I had Gemini uh actually run an audit on the findings, and we did find some discrepancies and errors. So again, that's super, super, super pro tip. You don't have to do that. But for me, and I'll show you here an example, what that can look like, but it really dictates how I function in my operating protocol for myself. Okay, so next from this project thread, after I've established this, what I want it to do, after I tell it kind of what I need you to be, what I want it to accomplish, I have it write instructions that we save at the project level. And the reason for that is because as you start to build more threads, you've probably realized that you go from one really, really optimized thread to starting a new one. And it kind of takes weeks to get back to that one that was really optimized. So, really what you want to think about is when you give it instructions or files or it's building memory, that's like the external hard drive for your project. It's like you're building the brain and you want to feed it really critical things that it needs to remember. So, like, think about when you're working with a client or you're doing something uh in in business or you're teaching something, there's like certain key takeaways you want people to have. So, like you'll summarize the end of a chapter and be like, these are the things to remember from that. These are type things that I pull from all of my threads and I save it as a project file, or I make sure that I give it like top line instructions. So I can read you what mine is. I actually typically have this initial thread that I start create the instructions for the project itself, and then I audit it. That way it kind of makes it specifically with Claude, it makes it very detailed and specific. And I find that to be very, very helpful. But this is what I'm trying to accomplish here after I feed it the original files, right? Underneath the historical context, business details, and then the genetic SNPs. Okay, next, ongoing dialogue. And this is really important because how you start your thread and what you do inside of this thread inside of Claude, it starts building memory. And you'll notice it'll say like use more, and then it'll start adding memory. So things that are important, uh, it'll start baking into the permanence layer of its memory. And so it's going to remember things about you. And sometimes it's kind of spooky because you're talking about something that it remembers something from weeks and weeks and weeks ago. So sometimes you can tell it, please remember this, and it'll bake that into that memory file. Uh, or typically what it'll do is it'll extract things on its own and start building that memory, and it'll use its own logical reasoning as to what goes there. But for me, I like to make sure that at least I'm orchestrating a little bit of that as we do ongoing dialogue, meaning I'm using that thread, and then I'm basically telling it, like, yes, make sure you remember that for a long period of time, right? Like this is something to permanently remember about what we've discussed. So for me, it was the comparison to the other five or six like entrepreneur types or entrepreneurs that uh we we recognize who are similar to me, and then we're pulling protocols from those specific entrepreneurs that we added to a file, but also it's built into that permanent memory as well. So really, really important and super, super helpful. Okay, and then finally, and this is something I don't think many people do. So I think you're gonna be way ahead if you do any of these things that I told you so far. But what I tend to do as well is as I'm having ongoing dialogue, if we have, you know, those moments you have those like really profound conversations with AI and you're like, oh my gosh, I need to take notes on this. This is so good. What you can do is say, I love the work we did today. Can we save this as a project file so we can refer back to it in the future? There's nothing worse than having some really profound work done with AI, specifically when you're talking about working on yourself and having an executive coach, when you come back and you're like, where was that? I can't remember. Now, Claude fortunately has the ability to look back into old threads and find things like this. You can also fork that data and give it to a new thread. Again, topic for another day. But for me, while it's ripe, I've found it to typically not pull as much context from it in hindsight as if I do it right there in the moment. So when you have those moments of like profound, meaningful takeaways, you really want to say, hey, I think we should add that to the project files. And then your claw typically will agree because that's what it's trained to do for the most part. Uh a little bit of a side note on that, in your instructions when you go back here, you can always ask it to give you truth over comfort. So if you want it to be more real with you and not just confirm your bias or tell you what it thinks it wants you to hear, you can do that. But again, rabbit hole. Okay, so project files. Yes. As I'm going, I typically will ask it to, okay, yes, we need to add that as a project file or update the memory or update the instructions, whatever you feel like is again discernment as a pilot. What is this really best used for? For me, project files are typically like if we come up with a new protocol that's gonna work for me, or if there's a major update, or my supplement protocol that it recommended. So this is the overall architecture that I recommend if you're gonna build something like an executive coach using Claude. Now, I think this will work. Like if you build a gem, uh, I found gems to be less effective than Claude. That might just be, I've spent more time on Claude. But anywhere you can build projects like this, you could probably achieve somewhat the same outcome. So, whatever model you feel most comfortable with, or whichever model you use the most, probably just go with that one. For me, Claude's precision, I caught the WordPress of AI. And the reason is because WordPress was SEO friendly because of the way it was built, because it was built to be a blogger site. And so it had attributes that ended up being search friendly. Well, Claude was created primarily for writing code, and that's how people are using it. And so precision and writing code and detail and not lacking a lot of models will just give you a short answer. Claude gives you a dictionary answer. Some people love that, I love that, some people hate that, but the thing is that it makes sure it's focused on precision. And for me, if I'm doing deep work like this, I want precision, not just something I want to hear, or a short answer without a lot of context. So it's a preference for me. But I have different use cases for each AI model. This is just the one that I prefer for Claude specifically in this particular use case. Okay, so let's go ahead and jump into the Claude side of things so you can see what this ends up looking like because this is really important. Um, so I name mine Clue. If you're a Tron fan, you'll know why I called it Clue. Uh hopefully it doesn't turn on me. But anyway, so you'll see here, these are all my threads that I call main, and I archive the ones that are older. So typically I think the I think it's like 200,000 tokens per thread where you start to kind of what experience what's called drift, which means it starts to kind of lose because it's too much context. It's like having like a six-hour conversation with a human. Like you're like, oh my gosh, it's been forever. Like this is the longest marathon conversation ever. So you kind of need to move those over. And that's a video I'll cover in another day. But these are archive main threads. Now, if I have a separate project, you'll see this in my Tarvis thread, which is my strategy partner that for my for our online business, the Business Lounge. Um, there's other projects, like we had a loungy project, which was our ClickUp 4.0 project, where basically I start a separate thread, but I forked that from our main thread. So I did the strategy thought stuff with the main thread, and then I had it fork over the instructions and the details to start a new thread so that had its own environment where that lived there. So if they're gonna be big, robust things you're doing, I recommend starting a new thread versus having this huge conversation inside your main thread that's outside of the normal context that you have inside your main thread. But again, a video for another day. So for me, that's what it looks like. You can see I was just in it because I was tweaking some things. I'm not gonna pull it up because there's, I mean, it's my executive coach, right? There's some privacy stuff in there. But you'll see here what I told you. So the memory is coming from the conversations that we've had. Uh, and this is actually was created by Claude with the ongoing conversations that we've had. The instructions, you're serving as Chris's executive coach, therapist and functional medicine assistant, maintain a direct, challenge-oriented tone. And there probably says truth over comfort as well, which is my preference for any AI models. I just tell me the truth as you see it or as you've discovered it, not just give me the sugar-coated answer, right? And then you'll see here I have six files currently. That's actually light for me, but these are lengthy, these are pretty lengthy files. So this is 1500 lines, like this executive courts coach fork data from the original thread that I had is this one here. I have my uh business strategy plan. Probably don't need that one there. I have migrated that over to the actual strategy thought partner one. Um, this is my genetic profile. You'll see here. It shows all my genetic SNPs that it's pulled and all the things to know about it. This has arrived at Confluence again with Gemini as well. You'll see literally like all of this, what it means, what it does, uh the SAM, the folic acid, the folate meth uh, I mean, it's methylfolate, like it's crazy, all the recommendations, and it actually gave me a supplement protocol as well, based off of that. So, just as an example, what I mean by that is I could literally say, Hey, I'm feeling this way today. Can you look at my uh genetic report and also what we established as far as my health context and give me some suggestions about food, exercise, or supplementation I can take today, feeling this way. Now, again, it is not a professional, and I would go to my funk medicine, but if I'm in one of those gaps and I'm having a rough morning or something, this is gonna remind me of well, let's trace this thing back and see what your genetics are speaking to, and maybe they can provide me with some reprieve where I can then eliminate some of that, those gaps until I find a true answer. So it's very, very helpful. Uh the pattern interrupt protocol. So this is covers one of my gaps or one of my quote weaknesses, blind spots. But basically, my pattern interrupt, if I'm falling into like more of like traditional patterns, which are not necessarily productive or fruitful for me based off my historical context, I have a pattern interrupt protocol that stops that and it retrains my brain if I follow the protocol to build a new neural pathway around a different focus on how I'm pursuing what I'm pursuing, right? My personal operating manual, this is kind of the overall arcing, like summarized version. And then you'll see here I mentioned it before my supplement protocol. So all these are the currents. As I'm having profound conversations in here, I will continue to uh add more files to this. And then as you're talking to your different threads, you can ask it to search files for certain documents and refer to that if you have a question about it specifically. So it basically allows it to go inside of that external hard drive, pull it out for you, analyze that document, and pick up right where you left off. You can also ask you to look back at previous threads to see previous conversations if you enable it to do so. And then now you're right back, hopefully as close to optimized as you can be. All right, guys. So that is how I built my executive coach using Claude. If you want more information on how to do this yourself, we're actually offering audits, you can check that out. We'll actually go with you on a Zoom call and explore how you can build something like this for yourself. Both myself and my team can help you build something like this for yourself. You can check that link out below. Or there's additional ways you can work with me and my team, specifically if you are a coach or service provider in the online space. All right, before you guys leave, don't forget to like this video, hit subscribe, and hit the bell for notifications, and drop me a comment below about what you want to hear next. See you guys in the next one.