SPEAKER_01 0:34

AI agents can sound like something designed for large companies with technical teams, complicated automations, and very large budgets. But agentic AI can be much simpler than that. In this episode, I am sharing part of a recent training where I explain how solopreneurs can begin using AI to complete multi-step work. You'll hear the difference between agentic and autonomous AI, how a skill, a task, and schedules work together, and how to decide which parts of your business are actually worth delegating to AI. You don't need to automate everything. You simply need to identify one recurring task that feels heavy, give AI clear instructions, and watch the magic happen. Let's begin this episode.

SPEAKER_00 1:44

Let's amplify your authority.

SPEAKER_01 1:48

AI Agents Made Simple and we are going to discover how solopreneurs can begin using agentic AI to get real work done. So AI is moving beyond the questions and answers now. What we were doing with something like ChatGBT when it first came out, we would ask it a question, it would get an answer, then we'd ask another question, then get another answer, and then we'd start working, right? But nowadays it's a little different. Now we set a goal. AI runs a process, it takes multiple actions or multiple steps, it delivers finished work, and then all we do is review it. So agentic and autonomous are two words that we need to be very clear because the change that we're seeing, the before and after, is because AI is capable of doing more things. So these two words get used interchangeably, but they're very, very different. So let's go over it step by step. Agentic AI, AI can work through multiple steps toward a goal. You set the destination, in other words, you tell it what you want done. AI handles the driving, but you're still in the car. AI takes over and does multiple steps, but we're still in the loop, so to speak, as they say. What we're seeing now is autonomous AI. So autonomous is a little different. AI operates with much greater independence. It has autonomy. Making and executing decisions with minimal or sometimes no human involvement. That is totally different. That is where we're hands off and it is doing a lot for us. The only hesitation I have with autonomous AI is that everything leading up to it has to be scrutinized and set up very, very clearly and correctly before you set something like this up because it could make decisions that you don't want it to make decisions. So we have to be very careful when we get to this level of autonomous AI. But there's a lot we can do with agentic AI. A lot. And you may not even want autonomous AI. You may not want it to do things for you. Okay, some of what you're doing already is probably agentic. You may not realize it, but there's a lot of features within the tools that we already have that have this agenc feel to it or agentic capabilities. We might not realize it, but it does. What you want to do when you're thinking of agentic is what work merits agentic AI. Look at all of the things that a solopreneur does. There's a lot from blogs, podcasts, websites, lead magnets, newsletters, webinars, YouTube, email, social media, case studies, sales pages, ad creative, repurposing. There's a lot of things that we do, and that's all considered marketing, right? Getting our word out. And there's a lot there. There's a lot of moving parts. And that's just maybe for one product that you sell. Maybe you sell courses and consulting, right? Those are two different things. So we have to figure out where can I get the most bang for my buck? What can I use agentic AI to do for me that will take a big weight off my shoulders and focus on those things? Because some of the other things maybe not be worthy of using agentic AI because it really doesn't take you that long. Or you can use another form of AI, like an AI chat or a project, to do it and not have to necessarily go agentic where it's multi-step. So think AI agents as if you were hiring someone. A lot of people use this comparison because it's it's easy to understand how we should communicate to AI. Because one of the things I realized when I started using AI is how poorly I communicated. I thought I was a communicator. Keep in mind, I was writing, I took public speaking, I thought I was a good communicator. And it wasn't until I started using AI that I realized, oh, I should have told it that. I didn't give it the context. So we need to make sure that we treat AI as if we were hiring someone, then we won't omit things, right? Before you write a single instruction for an agentic workflow, and we're talking workflows, we're going into workflows, right? Not just question-answer. Work backwards from the result. So what is it that you wanted to do? This is the same thinking you'd apply if you were bringing in a contractor or a specialist. Some of you are service providers and you have contract work. So think of it that way. Ask yourself in that order, right? What would I need to share? If any of you have hired a VA, you would do this as well. So what would great look like? You want to be able to communicate this is what a good output or a good result would look like, or a good finished product would look like, or a good sales page would look like, or a good graphic would look like, or a good uh funnel will look like. So you begin to figure out what does good look like? And then what professional would I hire to do that? Would I need a graphic designer? Would I need a copywriter? Would I need a strategist? What would I need? What do I need a business consultant? And you figure out the role, the type of professional that you would need. So there's a little bit of thinking that goes on first. And then what process would this specialist follow? Like what's the SOP, so to speak, the standard operating procedure for this? And then what would they need to know? Is there additional context? Is there maybe a brand kit? Is there something that they need to be able to understand? Not just the task, but also the macro view of it, how it fits all together in what you're trying to accomplish. So when we're thinking about a gen tick, we want to be thinking about these things so we can set it up for success, so to speak. So let's take a look at this. A skill, we're going to talk about what a skill is, is how you want the work done. This is where you have the specialist. How do you want the work done? A skill isn't just a prompt, it's a reusable set of instructions that captures how specific type of work should be done. Everyone has a certain way of doing their work. This is what you're trying to capture and share when you create a skill. You build it once and it works for you every time after that. So these skills are created, and these are the ones that know how to do the work. The task tells AI what needs to happen. So when you have a task, so you have an expert, copywriter, copywriter sitting on the bench doing nothing, and all of a sudden there's a task. Here's a task. What's the task? Is it a sales page? Is it an email sequence? Is it an opt-in page? Is it social media? What is it, right? Copywriters sitting there waiting. The task tells it what needs to happen. So a skill is a how a task is the what. So we have skill, how the work should be done. We have the task, what needs to get done, right? The specific job, right? Is it a sales page? Is it an email sequence? What is it? The inputs and expected output. So now you have how to do it. Then you tell it what needs to happen. And then a schedule task tells it when to do it. So this is when you begin to create that workflow that everybody salivates over, right? They want these workflows. When you marry the schedule task and skills, you'll be able to create some workflows that are manageable that will help you tremendously without overcomplicating everything. Here's where we're shifting from just having AI assistant to AI doing work for you. Adding a schedule to your skill and tasks combination means that the work starts and often finishes without you having to remember to begin it. You could set triggers for this, right? And part of that is scheduling a task. That would be the trigger. So everything I just shared with you in the last five minutes, we can summarize it this way: the skill is the how, the task is the what, and the schedule is the when. If you remember that, you'll be good to go. Agentic assistants begin doing the work for you. You review the results instead of managing every single step. All right. The trigger would be the schedule. Like if you say every Monday, I want you to do this, or every uh every day of the week from Monday through Friday, I want you to do this. There's a trigger. AI begins. It will use the skill if you tell it to. If you say every Monday, I want you to do some research, and then I want you to use the skill to write a blog post and then use my social media skill to create some social posts, it will do that. So in the task itself, you can tell it what you want it to do. It will complete the steps and then it will deliver the results. I like in Chat GPT, it actually sends you an email and said, hey, it's ready, so you can review it, which is pretty cool. So from research to ready to review content. Again, here's the example. Every Monday, instead of starting your content week from scratch, a scheduled agentic workflow handles the research, the drafting, and the repurposing and delivers everything to you before you finish your first cup of coffee. So the scheduled task is triggered every Monday morning. The workflow begins automatically. No reminder, nothing. It just does it on its own. And then it starts researching. I have one that researches YouTube videos. You can be very specific. What do you want it to research? If you have specific sites that you like to research, tell it. Tell it these are the sites I like you to research because maybe there's content in there that you really value and you want to stay in the loop of publications that they have. So it could go through the research and tell it how many you want it to research. Because if you just leave it open-ended, it's going to use a lot of credits or a lot of bandwidth there. It's going to work hard and you may not need that many. So just say, you know, I just want some really good relevant topics and give it the parameters. You know, this is for this type of audience and this is what I do. So it has to be relevant to me and how I serve my clients. And so it's going to begin to look. And then you can tell it to use the newsletter skill. If you created a skill to create your newsletter in the format that you want, in the voice, tone, and style, with all of that, the things that you include, the subheads, the way you like it structured. You can say, I want you to choose the best one, knowing me, and you give it a brand kit and you explain. I even give it my personal assessments. So it knows very much how I am. So I give it my personal assessments. I have a brand kit, it understands what I like, what I don't. Oftentimes AI will say, oh, when we're working, it'll say, This sounds like Marissa. I bet you'll like this one out of all these suggestions. And nine times out of 10, it's correct. It knows me that well. So you can have it begin to write based on the research it did. And then repurpose the content after it's done and take that newsletter content, repurpose it, and create some social posts for the week. Maybe you want three social posts and tell it what kind. Do you want a quote? Do you want a video script? Do you want just a post? Tell it what you want. Do you want to create a poll maybe on LinkedIn? And it will do that for you. And then it'll deliver for review. So with mine, it shows me the five that it researched and the one it picked. So I can look and see which out of the five, which one did it pick and what were the four that it didn't like that it it figured didn't align with me. So I can review it and see if it's if it's hitting 10 for 10. So then I can see the draft. And usually I add my own little story. I add some things, but I'm not starting from scratch. I'm not starting with a blank screen. And then I have the social post and maybe something, a quote. I like it to always pull a quote and write an original quote for me because I like quotes. And it already has my brand kit, so it's got some of my quotes in there. So it knows how I like to frame those. And all I have to do is review it. This is where it becomes very agentic. So Monday morning, rather than go and start researching stuff, I'm just reviewing things. Those are just a few ways. Now, the instructions still matter. Now, these are some of the common ways that people quote unquote prompt. So the first one is the structured prompting. You know, you give it a role, you say the task, you give it the context, you give it the guidelines, and you set the boundaries. That's still good. It depends on what you're doing, right? The other way is the natural prompting. Maybe you click that microphone and you just have a conversation and you explain your goal, you ask for follow-up questions, and that's how you begin to prompt. This is number two, this natural prompting where you just open up the microphone and you start talking to it, is one way you can create skills. The third is learn from success. Say you're already in a chat and you're going back and forth, and what you produced is amazing. You and the AI have produced something amazing. You say, you know what? I want to be able to do this the next time, just like we did right now. Can you create a skill for me? And that's another way you can do it. And then plan before you build. Like I said, you could, in the situation that I was talking about with the skills in the chat, you could start there and then take that information and go to the skill creator and say, This is what I have so far to create this skill. Or if you have another tool that's outside and maybe it creates apps, I use Lovable and I created a number of apps. I don't do a lot of the work in Lovable, otherwise, it'll eat up credits like crazy. I will go in ChatGPT or in Claude and I'll go through it there and I'll have it help me communicate what I want to see in the app in a way that the lovable platform will understand. Why not have AI write for AI, right? It makes it a lot easier because they understand each other. So before adding a step or before adding an agentic workflow, you can test it. And these platforms always have a place in the settings where you can see the usage. So what I do when I'm introducing a new skill that has multiple steps is I will look at the usage, I will go create it, I will do a test run, and I'll go back and I'll look at the usage to see how much it used. And that way it gives me an idea if this is something I can do long term within the allotted credits that I have already or the tokens that I have already. Or if I do need to buy more, you know, how much is that going to cost me? You don't need to become an AI expert. You don't have if technology is something that ugh just leaves a bad taste in your mouth, you you really don't. You just have to think through your work and figure out what would really help you and begin to communicate clearly, maybe to yourself first. Maybe communicate it to your husband or your wife. Do they understand what you're trying to accomplish? And then go to AI and say, hey, this is what I'm trying to accomplish. And it will help you. It will ask you questions. If you tell it, ask me questions. What did I leave out? It will help you with that. It's amazing what it'll do. So the solopreneur who gets the most from AI, they aren't the ones who know more about how technology works, all the back end bells and whistles. They're the ones who are clear about their own work, their own destination, the things that they're trying to do. If you don't understand like how your funnel is going to work, you can talk to AI and figure out what kind of expert do I need to improve my funnel. And you can begin to do maybe research and then ask AI to model a certain expert on funnels to be able to see and assess and analyze why your funnel isn't working. So if you can have AI be agentic and take one of the things that's really feels heavy for you to do week to week and really master that and get it cleaned up and get it set up with just simply two things, the schedule tasks and skills, because we have skills now in Chat GPT. We didn't have that before, we have it now. Then it's going to help you. And you're going to be able to see the magic of workflows just with those two features. And all of the AI platforms have those two features. You're going to be able to do so much. You're going to think, why was I thinking of using N8N? And why was I thinking of using Make and Zapier and everything else when this does so much? And with Chat GBT, it gives you a little notification in your email inbox. Ding, done. So with the others, you go there to the platform and you can see what it produced for you. Try the schedule task and the skills in whatever platform you're using. Start with one task, not a complicated automation. You can have a couple of steps that you want, like the kind I explained. You can get it to where it's done really well. You'll be able to see how well you communicate it with it. And pick something that's repetitive, something that feels heavy, something that would really help you in the long run, something that takes a long time for you to do. And start with that and wait to see the magic happen. It's pretty incredible. Pretty incredible. Learn a let AI earn the right to do more. Once it's done right, then you can get into larger workflows if that's what you want to do. So that's what I have for you. I so appreciate you guys just being here and listening. I really appreciate it. It's always an honor. I hope this was helpful. Take care and I will talk to you soon. Bye bye.