How to turn Claude Code into your own executive assistant (Game changer)

Claude Code is quite literally changing the way that me and many other people are running their businesses. I've been using it for about the last 6 weeks, and to be honest, it's one of the most excited I've been about any AI product since ChatGPT came out. So in this video, I'm going to be showing you exactly how I use it every single day to essentially be a one-man army. It's allowed me to do things way quicker than I previously would have done them, and it's also allowed me to not have to scale up my headcount in order to be able to do certain tasks. So if you stick around till the end of the video, I'll show you how you can essentially have your own AI operating system that can do pretty much everything. Thing for you. And I'll also show you how you can get the template that I use so you can get started with it straight away. But without further ado, let's just jump straight into it. In this video, I'm going to be showing you the 9 features that are going to take Claude Code from the basic version all the way up to a system that can do pretty much anything for you. So this is my personal Claude Code instance here. As you can see, it's just called Autonomously AI. It manages everything to do with my business. And the first thing, and possibly one of the most important things that I want to show you, is the context. If we come over to this context folder here and have a look inside of this, first of all, we have all of these MD files that basically just break down all of the core components of my business. So we have brand voice at the beginning here, and this just helps when it's doing anything like writing emails, writing LinkedIn posts, or YouTube captions, or anything like that for me. It's quite simple, but it basically just talks about how I myself, um, like to come across, uh, in writing. So I'm not going to go into too much detail here, but you can see like email style and then some examples, and then the YouTube video styles also helps with scripts YouTube videos and things like that. And then we have business information, and this is by far one of the most important out of all of the context files. So first of all, it just tells it how it connects to the other files that I'm going to show you. And then we just have an organization overview here, the different products and services I sell, the pricing that I usually do, and then also some key context. And then also just some extra information about website and brand. So my website here, this is my YouTube, and then these are my design systems. So things like the fonts, the primary colors, all of that kind of stuff. And then we have the clients.md folder. I'm not going to show you inside of it only because I don't want to expose all of my client information and project information, but in here it's just a detailed list of basically all of my active clients, um, who's the main contact, the deal size, if we've got the first upfront payment or if we're waiting on that second one, uh, the current stage of the project, everything like that. And then we have the current data.md that basically outlines like my current revenue, the current things that I'm trying um, you know, my strategies with YouTube, if I'm trying any other lead acquisition strategies, everything like that. And all of those kind of numbers and stats, that's what goes into Current Data. And then Personal Information, uh, basically says my role at the company, so what I'm doing, what I'm responsible for, and how this workspace is going to help me. And then Strategy. So basically, this is the things that we're kind of like focusing on at the moment. So we've got the current focus period, the strategic priorities, and then also what success would look like. And then we also have Team, which kind of looks like Clients, where it's basically a detailed breakdown of everyone on my team. So all of the developers, sales team, everything like that is all in there with a detailed breakdown of all of their contact information, the projects they're working on at the moment, all their responsibilities, um, you know, if I'm paying them on like retainers, how that looks. So that is all the context files in there. I also have a few extra ones in here, things like proposal examples, so examples of proposals I've done so it can help me write proposals. And then also brand assets in here, it's just, um, logos and stuff that it can use when it's generating PDFs or just for reference. So this context file alone is by far one of the most important parts of it because it kind of ties everything together and makes it so it's not just like an external assistant that I'm having to like constantly remind who I am, what I'm doing. I can just say things to it way more naturally and just say, oh, can you, you know, check up on this person and see if they're still doing this? Or I can talk to them about a client. I don't have to give them a full background on what we're doing for that client, who's the contact. I can just talk to them about that client Uh, say I've got an issue with one of their projects, I can just talk to it about that client. Instantly knows what I'm talking about, what the project is, um, you know, the contact information for that client, where to get any extra information on that client, everything like that. It makes it absolutely amazing. And then the next thing I want to show you is the two commands that I use every day. So one of the cool things about Claude Code is you can have multiple tabs running at the same time. So every morning I log in and then I have two tabs open, and the first command that I run here is /prime. And basically what this does is it just primes the project here, or the instance, and it gets Claude Code to just look through all those context files, sees what I'm currently working on. Um, if it needs to update any of the client information on the team information, it will just go ahead and do that and basically just get it ready to start working with. So I'll go prime like this, and then in the other tab here, what I do is /brief. And what Brief does is it basically looks at all the data from yesterday. It's connected to my Stripe, my Gmail, my calendar, everything, sees what I did yesterday, and then gives me a bunch of like action items, things that I need to do straight away. Reply to this email, send this person this document, everything like that. So we'll send that off as well. Okay, awesome. So first of all, it's given me the priority actions here, which are just things that I basically need to do straight away today. And it's given me today's schedule, and it's given me like an overview of my pipeline, so all of my active projects. And then the things that are like pre-close, so leads that I've been speaking to. And then we have content and brand that just gives me an update on like my current subscribers, the amount of views, the videos published, and also my latest videos. And then just a quick summary of my inbox here as well. Okay, cool. And then Prime is just finishing up here. You can see it knows who I am now. It knows everything about the workspace, knows all its commands available. It gives me an overview of the strategies and priorities, all of my active clients. And then one of the projects I'm working on as well there. So awesome. I do that every morning, and that just gets one chat, um, and that just gets one conversation ready to start working on other stuff. And then it also gets one conversation that just gives me, uh, basically an output of the things I need to do every morning. But while we're on commands, I want to show you the other commands that we have. If we go to Claude here and we go to commands, so Brief, I've already showed you that. Content Ideas here, this is essentially when I'm coming up for content ideas with YouTube, I run a slash command to give me content ideas, and it already knows all of my competitors, the type of videos I make, everything like that. It gives me some really great outputs. And then create a plan. This will actually just do exactly what it says. So if I want to plan something before, it can just create a plan by using this slash command. Draft an email. If I want to draft an email to a certain person, I just start it with draft email, then it instantly knows how I draft emails, how I sign them off. Um, and it can also know to research who it's going to in any prior context. And then implement. This is just when it's created a plan, I just run implement to actually implement that plan. Meeting Prep. This one is amazing. What it does, basically, I run Meeting Prep and it looks at all of my meetings on my calendar. If I say a specific one to it, it'll look at that one, but if not, it grabs all of them. But most of the time I say Meeting Prep, I've got a meeting with this new inquiry or this new lead or something, and then what it will do is it will look at their inquiry, so how they inquired, what they said in it, and then it'll look at their company, where they're located, everything like that, give me a full background on them, so exactly what they do, their company, you know, uh, their company's revenue, headcount, things like that. And also look at their inquiry and give me some really good talking points for the call and kind of how to structure the call. It's absolutely amazing. I do it all the time. And then pipeline, I run that, it just gives me an overview of my pipeline. Prime, you've already seen that. Proposal, when I run that, it can generate me proposals. So if I say proposal, generate me a proposal for, and then maybe, you know, a lead I just had a call with, be able to, it knows with this slash command that has to look at my calendar, grab the transcription from the call, look at my past proposals, see how they're structured. And then give me an output like that. And then Queue is just going to surface basically any outstanding tasks that I have that I need to do. So that's really helpful. Sync Clients— this one's really cool. Uh, when I run Sync Clients, it's just going to go through my HubSpot, my email, uh, all of my interactions, my Google Meets, everything like that, and basically just sync all the clients. If I've got new clients, it'll just automatically add them to those client MD. And if the projects move along in any way, or let's say they pay or it goes to like a next stage in the project it'll just add all of that into the MD file so it's constantly up to date. And then Sync Data, which is basically like a lighter weight version of Sync Client that just does this for all of my data. So if I want a quick snapshot of basically just everything in an MD file, like all of my current deals, my current team, um, everything I'm currently working on, I just quickly run Sync Data and it'll do that for me. Okay, so now we move on to tools, specifically the MCP connections we have. And this is what ultimately takes it from just a regular chatbot that can give you text outputs do things that can actually take action for you and go externally outside of Claude Code. So I have two MCP servers because that's really all you need in my opinion, at least for the basic stuff anyway. If I go to manage MCP servers here, you can see the two ones I have is the NAN MCP and then Rube. So with the NAN MCP, basically this is connected up to my NAN instance. So if I want to have some kind of automation inside of NAN, or I can also connect up to client accounts so it can help with debugging and things like that. This is awesome because it's just a standard NAN MCP, but it's able to take action in there, put nodes down, look at the executions, chase things around, add credentials, anything like that. So this one is super, super handy. If I say to it, for example, I want a quick automation that when I get a new email coming through, it does XYZ, if it thinks it'll be better to do that in NAN, it can just go ahead and create that inside of my NAN instance. It already has all of my credentials pre-filled. And then we have Rube, which is arguably the backbone of all of this. And basically what Rube is— I did, uh, make a video setting this all up and showing you around it, so I'll link that video. But essentially what Rube is, is one MCP that connects to all of my MCPs. So through Rube, I have things like, um, Gmail, Google Calendar, Google Drive, Google Docs, Google Sheets. I have YouTube, Canva, Stripe, pretty much everything I use, all connected. Their MCPs are connected through Rube, and then I just connect Rube into Claude Code and has access to everything. It's absolutely amazing. Like I said, I'm going to link the video where I set that all up but this is an absolute game changer because it means it can read anything from my calendar, from my emails, um, it can look at my Stripe, but it can also take action. So it could create me a Google Doc, it can email people back, it can create a new sheet, it can do anything. And then now the next part I want to move on to is the agents that I have. So I don't have many agents, I am working on creating some more, but basically these are like sub-agents that are for me just specialists in certain things at Claude Code can call on to, you know, collaborate on tasks or just hand over tasks to it. So the first one I have on there is a YouTube Strategist. So this basically knows everything about my YouTube, all about my competitors, everything like that. And it's really, really good when it's coming up with video ideas or helping me plan a thumbnail or an intro or a script, anything like that. A lot of the time what Claude Code will do, if it's a YouTube-related task, it'll just pass this over to the YouTube Strategist, which is just way better it's pretty much good at just YouTube stuff. So it knows everything about my YouTube and YouTube in general, and I use that all the time for YouTube-related tasks. And then similar to that, we have an NAN specialist here because NAN is obviously quite a big platform and their MCP is pretty massive. There's a lot of things to, um, you know, a lot of moving parts in there. This agent here is just an expert with the NAN MCP. So when I want to do something in NAN, Claude Code will just pass it over or collaborate with this specialist agent here and it just knows everything about setting up nodes in NAN and, you know, looking at executions. Because sometimes when you use the NAN MCP, um, Claude Code might, you know, put the metadata wrong in for a, um, certain node, or, you know, misspell something in the JSON, or use an older version or something like that. Things like that just break the whole system. It has to be really, really precise. So this specialist agent is just really, really good at that. So for example, I could say something like, use the YouTube Strategist agent to help me come up with 5 video ideas. We'll send this off. Then yeah, you'll be able to see n8nspecialist.md. So it's using it off the bat, which is amazing. Spawning the YouTube strategist now to pull live channel data, check competitors, and generate ideas. So it's also going to use the Rubemcp here because it's connected up to my YouTube. So we'll let this run and I'll just show you the outputs. Okay, so it just finished. You can see it did quite a lot of work here and it just gave me a channel snapshot at the top and then gave me some competitor intel. And then the top 5 videos: I let Claude Code build my entire NAM workflow library, here's what happened, the fit is high. And then Claude Code agent teams change how I run my AI agency, how I built a WhatsApp AI agent that books meetings for my clients, self-healing NAM workflows, and then 5 AI automations to make me $15K a month. And it gave me a full output here, as you can see. So that was the YouTube agent in action. It's really, really good, knows all of my competitors, knows exactly what to search, knows everything about my channel, all the videos I've done well and the videos I haven't. Okay, and now we're moving on to hooks. Basically, hooks in Claude Code are just Claude Code doing something at a certain time. So the hook being, um, the event. So it could be on session end when it's finished doing a task, when it started a new task, every 30 minutes, anything like that. Those things are hooks. So I only have one set up right now, and that looks something like this. So basically what this does, I think it's really, really nice because what I do with Claude Code is I'll let it run and then I'll go back over to like Google or WhatsApp, whatever I was doing, and then I'll come back when it's done. And what this does is it just sends me a notification in the top right corner of my screen when a task is finished. So I'm just going to say to it— so I said, can you just demo out that hook basically for me? Um, and then we should see it activate that. And then this just happens, yeah, when a message finishes. So yeah, I actually forgot, I, um, I didn't change it. I actually get a little pop-up that looks like this that just says Claude Code task complete. Claude is waiting for your input. And then the next thing I want to show you inside of here is we've looked at the agents, we've looked at the commands, and now I want to look at these skills. So First of all, we have the Skill Creator skill, and this just means basically Claude Code is easily able to create skills for me. So if I, you know, we do something good that's really useful that I want to do all the time, I'll say to it, create a skill for this and become an expert on it, and then it will create this skill so I can just easily invoke that whenever I want. And then we have the MCP integration tool. So when I want to add a new MCP, it uses this skill that basically makes it an expert on how to integrate MCPs and it allows us to just do everything on its own. So if it needs to do an MCP, we need to go out onto GitHub and find it and install it, yada yada. You just do that all by itself using this skill. And then we have Front End Design. This is when I'm building any kind of like mini apps or anything that's just going to be, you know, client-facing in some kind of way, or, you know, something that I'm just looking at. Just a really nice little skill to have. It just makes your front ends look like easily 10 times better. It just makes it an expert at making front ends. And then we have Create Recipe, which is actually really interesting. This is kind of like two things in one here. So Create Recipe, what this does— recipes are these things on Rube that basically just allow you to set automated tasks. So I'm going to show you inside of Rube what that actually means. So if we come over to Rube here, this is the same place where I have all of my tools collected up. As you can see, if we go to Recipes, I have one which is a new inquiry intake. So what this does, every day at 11 PM it goes from my inbox, see if I've got to know any inquiries, and it just simply adds them into HubSpot for me. So quite a simple one, but this just allows me to create scheduled tasks, and these run for free in Rube. They're just basically Python scripts that run. And what Claude Code will do when I invoke this skill and I tell it what, um, you know, automated task I want to run— it's usually just like really simple tasks that I just want running on a schedule forever— um, basically what it does is it creates the Python script first, it will test it out and basically debug it and make sure it's all working, and then it can just upload that into Rube create it as a permanent automation inside of there. So really, really awesome. Just little things that I want to automate, you know, make it so every day at 8 AM I get an email with the latest AI news or stuff like that. You can just do that through Roup. And then last but not least, we have the Claude MD here. And basically what this MD file does is it's acting as the brain for our Claude code here. The main thing this does is essentially just orchestrates it and tells it exactly where everything is and how to find everything. Everything. So it gives it a workspace structure here of all of our commands, our agents, and the projects, all of the context files here that, you know, the current projects I'm working on, the plans, the outputs, everything like that. It gives it like a breakdown of the structure so it doesn't have to look at the structure. It can just look in here and see exactly where everything is. And then it also gives it the key directory. So context, what is in context, uh, the agents, what's in the agent section, uh, plans, the different commands it has here. So basically just like the key things it's probably going to be using a lot. It just has some kind of like quick actions in here, so it doesn't even need to look through, uh, VS Code. You can just look in here. And then we have the Rube MCP, a breakdown, because I use that all the time. It's really important. So we have that in here. And then also the NAN MCP, because I only have 2 MCPs, makes sense to just put them in here. And then the hooks that we have, the sub-agents in here, the commands, which is Prime, which I just showed you, Brief, sync clients, pipeline, and all the other commands I've already showed you. And then the critical instruction: maintain this file. So whenever Claude makes changes to the workspace, Claude must consider whether the Claude.md needs updating. So basically, this makes it so everything is synced in the Claude.md. So Claude just has to look at this file to be able to quickly know what it has to do. So if I, for example, ask it something that requires some kind of context and I haven't done the prime command, it's going to look at the Claude MD, know where the context files lie and which one it probably needs to look at, and then instantly go there and read that. So it makes it way quicker. This is basically just Claude's second brain that tells it where everything is and how to do everything from a brief overview. And then what Claude does is it reads the Claude MD to know what it has to do, and it will go into these other files for commands or agents or context, um, it'll just be able to go straight there to get the information that it needs. So guys, that is it for the video. As I mentioned, you can grab the template for this. I'm going to leave a templated version in the description of this video. So you can just come in, install it into VS Code with Cloud Code installed, and you'll just have to fill in all of those context files and create your own agents and commands and things like that. But a lot of them will already be preloaded. Um, you know, there's commands and skills that are really specific to me, so they won't be in there, but about 80% of this will be templated, so you can just get started with it straight away. So that's it for the video, guys. I hope you enjoyed watching it. Make sure to grab that free template, and I'll see you in the next one.