You're Wasting AI on the Wrong 90% of Your Business

If you're using AI in your business right now, there's a 90% chance you're stuck at level one and you don't even know it yet. But if you're somebody who's never implemented AI into their business and you're curious about where to start, that's an even better reason to continue watching this video. And I know most of you are just using Chat GBT right now for simple conversations. And the biggest problem for you guys using chat GBT for your business is the fact that the memory gets filled up very fast. And every conversation that you start with chat starts from scratch. And the real play and using AI for your business is setting it up so that your AI agent evolves over time specifically to your business. And you, as you can see here on the right of my screen, I have all of my business files readily available for Claude Code to read, and they're easily readable by my AI. So every conversation that I'm having is not starting from scratch. It's starting from a place of awareness where my KPIs are, what my SOPs are, and the direction that I'm heading with my business. You don't have to be a coding expert. You don't need a technical background. I'll tell you right now, your business is definitely not niche enough for this setup to not work. So, in this video, I'm going to lay out the five levels of using AI inside of your business, and I'll show you some implementation so you guys can see and discern for yourself how to properly apply AI for your business and throw that setup paralysis right out the window. So, I showed you guys briefly what the five levels were before, but here's the TLDDR so you guys can get the full scope before we go through each level, some implementation, and main takeaways for you guys. Level one and level two consist of people who are mainly using it to augment some of their decisions and get some data. Basically, a quicker Google. You're asking something to chat GBT. You're getting the data. You're copying, you're pasting it, you're putting it somewhere else. And the lie that stands still at level one is that AI is just a chatbot. And the worst part about being stuck at level one or level two is the fact that every conversation is starting from zero. Doesn't know who you are, your business, or your docs. And here's a perfect example to show you that there's a better way. This exact image was just created through Claude Code right here in my setup for my brand Charlie Automates. I have my brand context kit right here with the fonts I use, the colors I use, my ICP, and I have a connecting software directly here inside of my setup called Higsfield. So, I can grab that business context, have a conversation, send a prompt over to the AI image generation tool, and pull that right into my system. And I can ask it specific questions about this asset, how it fits into my other business, my school community, and overall my other products. And this is my agent brain here, right? So, I'm sitting at level five. I have all my assets inside the file system that you see here. I have them scaffolded out on a brain, and there's relationships that are tied in between all the assets. So, my clawed code understands what everything means to one another. So when I ask it a specific question about a deliverable that it created here, it gives me key insights and how it fits all together with everything that I have available and going on within my business. And I know I just said a lot, but that was a blend of AI augmentation, AI automation, and then I had the discernment to be able to go through, connect all the pieces, and see how I'm truly going to be using this for my business. And this showcase image that I created with my setup is truly the tip of the iceberg when it comes to using AI. And now you may see all my files and think it looks super complicated, but I promise you it's really not. Because at level two, you're figuring out how to organize all your information, how Claude can read everything. And the biggest leap that you guys are going to see in this video is going from level one to level two, which is the organizer. This is the real difference between the people that are getting results and just slop. What we're looking at is the AI brand that I'd mentioned. If you guys want to learn how to set this up, I used a plugin called Graphify. And this is Obsidian. I have another video that I did on this. It's either going to be linked above or in the description below. But the way that the information is laid out and how I have it structured here in just general files, is how I'm able to get consistent marketing output, email copy, automation of daily tasks, and something I know a lot of you guys are big on, social media automation, as well as lead generation. And I had mentioned real quickly that I was using Claude Code. I have a full beginner tutorial for non-coders on my channel as well. I'll also link that in the description. And I'm using Claude Code inside of VS Code. So check out that video if you're not ready to move forward yet. So at level two, becoming the organizer. Huge leap. And there's two parts to this. The file structure as well as the frameworks that you're using with your AI. So your file structure is how Claude understands your business, you everything, all the details. But the frameworks is how you're able to get sick output for everything you're looking to do for your business. Whether it's just a PDF you're trying to generate, a new automation you're putting together, onboarding sequences, lead intake. What I'm showing you is almost 2 years in the making using Claude to get my proper outputs and scale my business more effectively. I have videos on all these other tools aside procedure ops right now. But Paul is basically how I'm able to build my applications, build my websites, build like a professional engineer without ever even being one or going to school for it. Base is how I'm able to maintain the state of my workspace, all my files, details about me. Carl is another knowledge layer helping me organize all the information. So these two together is a match made in heaven. Seed helps me put all my ideas together. I feed that guy right back over to Paul and I'm able to build with proper ideas templated that the AI can read. So my output with AI is naturally 20 times, if not a 100 times better than the next guy using AI without the proper file structure or frameworks in place. And I've productized everything that you see here into something called Charlie OS. And if you guys want to work with me on this, the first link in the description is how you guys can apply and see if Charlie OS is going to be a good fit for your business. Also, a quick segue. If you're getting value out of this so far, hit the like button and subscribe for more content like this. I put out about one to two videos like this every single week on AI automation, AI augmentation, and I'm biased. I do a lot of my content on cloud code. But back to the main topic here. I just want to give you a highle understanding of the rest of what we're looking at here. Then we can move forward into level three. Skillmith, that's how I'm able to build custom workflows tied to all of my external softwares. And they're keyword activated. So, as I'm having a conversation with Claude, like how I told it to create the image here, I created the skills and workflows with SkillSmith to tie in the Higsfield to automatically know where to go to generate that image or video. And Aegis is next level. It's a codebase auditor primarily for code errors or security vulnerabilities. And I just recently added procedure ops. It's my SOP engine. Last but not least, I did mention Graphify. It's my agent brain. So when you tie all of this together, vanilla chat GBT versus my file structure with all my data, essentially my evolving AI match with my frameworks, you can see how there's a clear difference between the majority of people using AI and the people who are truly excelling with it. All right, ladies and gents, let's move on beyond augmentation. And it leads us into the realm of automation because everybody believes they should start with automation. They want to automate their sales. They want to automate visits to their grandparents. Heck, I've even seen people automate messaging with AI to their spouse. That is nutty. I don't recommend it. And I'm newly wed, so I can't afford that right now. But we're looking at the builder and the orchestrator. And when we move beyond level two, you're crossing the Savant line because now you're not only using it to augment your life, but now you're creating tangible products that automate tasks for you, other digital assets like landing pages, websites, client portals, sellable SAS products. And I hear a lot of people say that they can't trust AI to do this one thing. Me being at level five, you need to have the discernment to know what to hand off to an AI, what to hand off to one of your employees, and what you should be personally handling. And maybe it's a task that you could just use AI to augment the data with your final call and a quality check. We'll get into what that looks like, but at levels three and four, we're going to talk about the implementation, what you can use to start building and orchestrating some of these solutions. And I'm going to tell you right now, this is how I think about building solutions. And this is how I'm able to have trust in what I build. Find your top bottlenecks for your business. What's costing you the most time and money? Out of those bottlenecks, what is the highest lever that we can pull that'll give us time back and make us more money? And do a dry run. Create an automation using AI. I'll also attach some of the videos that I have on my channel that I'll show you how to do that. Make sure it's a low stake automation. And then just watch it run. Do it a few times until you can see the output is consistent and that you can trust it because the reality is most people are stuck beyond this line here. But real leverage is starting above level two. So the cheat code is turning your most repeated task into an automation or a skill that can run autonomously that you've ran and dry run a few times so that you can build that trust because as an artist for example or a chef, your food or your art is only as good as the tools that you're using and the brain and skills behind it. And that leads me into my custom skills here within Charlie OS. And most of you guys believe this is just developer stuff. But that's a massive lie. These custom skills are literally just text files or as AI would understand it, a markdown file. For example, this plugin right here, SkillSmith, helped me create something called SOP build. And that's actually procedure ops right here. So my system is actually creating better pieces of my system with the pre-existing frameworks already in it. and it scaffolds out the file system for the departments of your business. This particular skill that I created is an interview process and it'll go through and ask me what I want to create an SOP for. Maybe I want to audit an existing SOP which goes back into augmentation. Maybe I want to give it a loom recording that it can transcribe and create an SOP from that video. Heck, even the conversation that I was just having with it, which I did a lot of chatting with today, I can create an SOP for. Or maybe we just want to start from scratch. three hours of pondering on SOPs and going through data versus a five minute interview feeding it all the data getting the output and then just auditing that for maybe another 15 minutes. I'll take the AI automation as a winner on that one. Or how about bottleneck pipeline? I use skillsmith to create this one for me. I'm using go high level and this is one of my pipelines available and you can see I have many stages where my leads are going to go through and I have to organize them and reach out accordingly. Right? I created this particular command stacked with a few different skills to be able to audit that pipeline and route those fresh leads to the right stage and automate the outbound communication based on the reports that they're filling out. So, if I wanted to look at the new reports, I would just tell it one and then it'll go through these new reports for me. I'm not going to share their information, but I actually have them fill out a quick quiz that'll go through, find their bottlenecks, give them clarity, and also give me clarity on what they're dealing with. And again, this was also built with clawed code using the seed and Paul framework. You see, things start to click yet. Use my brand kit. Here's the idea. Give it to seed. Let it package it up. Send that guy over to Paul and create the digital asset for me. But we could see here it went through, it found the three leads. It called my go highle MCP. MCP is just an acronym for a connector between Claude Code your AI and another software that you're using. But it gave me the full rundown, went through their details, and told me if they were high ticket or low ticket leads. And then I can go on and say customize an email, send them my booking link, or send them my landing page. Again, landing page, website, all built with quad code, seed, and poll. Now, I'm sure at this point beyond level two, you're seeing a huge leap into level three. But what does the orchestrator actually mean? What's the difference between the builder, the orchestrator, and ultimately the architect? And I mentioned it briefly before, and it's the MCP connectors. But some of these softwares may be familiar to you guys. Maybe you're using Slack, Trello as a pipeline. Maybe you have a school community. I know a lot of you are using ReadAI for your video transcriptions. And I use the GoHigh level MCP connector every single day for my business. So orchestration is a blend of augmentation. You're building some automations, but now you're orchestrating it all at level four and connecting all the dots and creating some of these streamlined automations for yourself so that you can save time. And it's not just your brain going through these processes. The frameworks and knowledge you've given it also help you audit what you're building to make sure that it even makes sense. I know it's hard to see, but procedure ops right now is basically my COO in a box. So with the right frameworks in place, the right custom skills that I've built using the frameworks inside of my OS, I can pull the strings at the top and automate pretty much anything. So if I have an idea, I mentioned seed, I can run seed and it'll ask me, are you building an application, workflow, campaign, etc. And after I've given it the idea, it'll tell me, do you want to launch it into a poll file? And these are the poll commands here. Let me take it a step further and visualize this for you. This is my Charlie Automate Websites file. I use Paul to help me build it. It helped me phase out the build. And I'm going to zoom out here so you guys can see this more clearly, but it actually creates the state files for the project. It's one of the benefits of using Paul. It's a unification system. So while you're building something with AI, it has a clear idea of where you are, what the project is about, and ultimately where you're going with the build of the website, application, etc. And that's another lane inside of orchestration that you get to go down once you have the right frameworks. So basically what we're looking at is a full command center using AI, not just general helpful AI conversations that most of you guys are limiting yourself at. So let me come full circle here. As an orchestrator, you use everything from level one to level three, but now you're connecting all your AI assets together through the MCP or something they call CLI tools. And what we're seeing here is a CLI, a command line interface. And don't let it scare you. And you can connect anything via an MCP or CLI tool. They're just bridges between AI and your software. And here's the G-code. Map out your departments, use the right frameworks, and wire the highest ROI tool first to your AI via MCP or CLI tool. And I would recommend using Claude Code. Again, tutorial in the description below. Could be a CRM, email tool, or just general analytics and let an agent own a cross tool workflow for you. In my personal OS that's available to my clients, I have my orchestration layers here. And I also have autonomous cloud-based agents, they're called Claude managed agents, here for me as well. These are running real business operations for me every single day without me having to go in, say anything, or do anything. So, my departments are running on Rails. This is easily well over $100,000 worth of systems work right here. And you can do it directly from your MacBook today. I won't beat a dead horse, but file structure and frameworks matched with, drum roll please, discernment. At level five, we've hit the architect right here. This is the master skill. This is where you stop trying to automate everything because it seems like a good idea. somebody said something to you or maybe you just saw a video. This is where you start triaging everything in your business based off of ROI. As I would call it, getting above the dance floor, looking at your departments where time and money are leaking and seeing where AI will actually play a part in this once you have the understanding and the skills available and frameworks to automate and augment your life and your business. To keep it simple, the four buckets are as such. Deterministic meaning no AI. Second one AI automation. third, AI augmentation, and fourth, human only. And my system has everything already put together. So when I asked it to put this asset together for me, it knew exactly what I was talking about. But let's talk about when to use any one of these four buckets. You would go deterministic, meaning A to B to C, when you need specific rules, high volume needs to be consistent and reliable. And the perfect example is a form that goes to your CRM with a template email. And you could use NEN, Zapier, cloud managed agents, but this bucket is where majority of your automations are actually going to go and they're typically the most fruitful. The second one is AI automation. So something that's repeatable, judgment, light, removable from you, and that could be like you saw the bottleneck pipeline, lead triage, weekly competitor briefs, outreach/ campaign drafts, the augmentation side of things. When you're going to use this is when you want your taste and judgment in the loop. We're looking at examples like strategy, creative direction, high stakes copy. You could take examples from previous winners on your own data or data from your competitors, feed that into your system with everything we spoke about from level one to level four and get solid output that you can use as a base and save yourself so much time writing from scratch. And at level five, the discernment is really key in understanding what is going to require a human only. where we're talking about dealing with people for the trust, the accountability. You're not going to have your AI agent focused on selling five, 20, $30,000 high ticket products. You're not going to have it being the sole decision maker for hiring an employee or for any partnerships that you're going to make for your business. So, inside of my business CRM, if somebody books a call with me, that's a very deterministic workflow and automation that I can set up that'll tell them when their appointment's coming, when they're confirmed, let me know, text me that I have a booking coming up, and go through touch points of some of the assets I need them to review before our call to make sure the whole process is seamless, and it's really straightforward top to bottom. Another deterministic workflow could be literally just having somebody opt in for a lead magnet or a free resource that you're giving them. You just send them emails and texts that you want them to see to create touch points. And this involves no AI. It's just how you speak, what you want to put in, that goes out automatically. What we saw with the bottleneck pipeline, when we're talking about AI automation, this is doing some lead triage for me. And when I have it review the leads in my pipeline, it's also creating drafts for me in terms of the emails that I want to send out. And that's also landing in a bucket of AI augmentation as well because it's creating those drafts and I'm able to go through it, review them, do a quick QC check, and then decide if I want to send that or change it up. And that's literally just a slash command that I have in here, but I can turn that into a daily automation that runs on its own. It sends me an email and says, "Hey, what are we going to do with this?" Always having a human in the loop, especially for things like sales. It's very important. When we move over to the AI augmentation bucket, I have something called thumbnail packager. So when I create a video, I can feed it an entire transcript, use best principles, and it can give me a framework or some ideas of how my thumbnails should look for my videos. And I have a file here. So I did a discovery on YouTube using another framework inside of Charlie OS to create a competitor's report. It looked through the top thumbnails and titles recently, and I can feed that into context here. So it'll help me augment my decisions of how my packaging should be on my videos for my titling and my thumbnails. And as you saw, I have the MCP for Higsfield. I can send it directly to Higsfield using all this compounded setup and create sick thumbnails that I can put for my YouTube videos. And the last bucket, human in the loop. Okay, when it comes to things like phone calls, you could have an AI agent to set up appointments and do some sort of qualification, but when it comes to humanto human connection for high ticket sales, phone calls, Zoom calls, that's you, baby. That's always going to be you. until further notice. I guess even when it comes to reviewing applicants that I want to hire for my business potentially, I can use AI to look through some of that data, augment my decision based off of who they are, what they've put in their report or their actual resume, but that's still my decision to make, not the AIs. The biggest trap that I see business owners falling into recently is the fact that they think everything falls into the AI automation bucket. But most of the automation opportunities are going to fall in the deterministic bucket or grinding on things manually that actually should have been deterministic or AI automated. And that comes with discernment. And most business owners have the skills to triage and figure out what's better to be handed off to an employee versus them handling it themselves, etc., etc. So, we went over a lot here and I hope you guys got some value out of this. Make sure to like and subscribe if you like this content. And if you guys want to work with me and build out this department of your business, just click the first link in the description to see if Charlie OS would be a good fit for your business. Watch the video, fill out the application, and we'll see if we can make something work. I hope you don't feel like I was mansplaining to you the whole time. This is just my perspective to you with my almost 3 years of using AI for my business, and I hope I get to see you guys on the next one. Thanks for watching.