AI Agents Beyond ChatGPT — Step-by-Step Guide & Transcript

[00:02] still using ChatGPT like a fancy Google search, you're about to get left behind. Most people are still chatting with AI, asking it questions, getting answers, maybe writing some emails. But the people who are going to win the next

[00:15] decade, they're not chatting with AI anymore. They're putting AI to work. AI agents that actually do things, build things, complete entire projects while the human sleeps. I build AI companies and tools every day. And let me tell

[00:30] you, this might be the most important technology shift we've ever seen. Bigger than ChatGPT when it came out, bigger than the internet. So, today I'm going to show you what this new AI actually is and why it's different, how to think

[00:43] about it differently so you actually use it right, the perfect [music] tool for your scenario, not just some hyped-up crap. And if you stick around to the end, I'll walk you through your first fully agentic workflow step-by-step.

[00:57] So, let's start by understanding what it is in the first place. Point one, what is agentic AI? There's three levels of AI. The first level that we all get introduced to was chat. ChatGPT, Claude, Gemini. I talk to

[01:12] it, I give it context, it replies. It's gotten better over time. The second level is automation. And that's when we start taking the AI and its ability to classify and respond, create workflows that run automatically. And these are

[01:25] tools like make.com or Zapier or n8n for the more advanced people. Level three is agent. This is where the AI thinks, plans, reasons, and executes entire plans, reasons, and executes entire tasks on its own. It opens browsers, it

[01:40] writes code, it creates files, it does research. It doesn't need your hand-holding. Most people are at level one, maybe level two. But the right the best part, you don't have to go to level two to get to level three because

[01:53] it'll do it for you. Point number two, the new mindset you need. Here's the biggest mistake that people are making with AI right now. Truth is they're barely using it. At best, they might be using it as a thought partner when it

[02:06] can actually build for you. This mindset shift that I see that I encourage people to consider, which is you are the director, not the doer. You need to be designing, not the task maker. And people think, "Well, if I don't do it,

[02:19] who's going to do it?" AI will do it. Your job is to give direction, to review the work, to course correct, not to do the work yourself, which sounds crazy. Essentially, you start with the outcome. What is the goal? And then you let AI

[02:34] If you're influencing it, then you're telling it what to do, and I'm telling you it knows better than you. So, if you haven't heard, IBM rolled out AI agents for their 270,000 employees. It was a $4.5 billion

[02:49] in productivity gains. Managers now complete tasks like managing their team and figuring out who needs a promotion 75% faster. That's not because AI is smarter than them, it's because they stopped doing and started directing. So,

[03:02] how do you actually become a good AI director? There's only three things you to have a clear outcome. You need to know what's the problem to solve. Before of things, you got to ask yourself, "What is that result? What is that

[03:17] thing is we want to give it clear instructions. I think about what is an ask it to do a task, can I give it a Can I give it the template? The more specific you can be, the better it's

[03:32] going to work for you. And three, clarify the results. I want you to treat it like an intern. Give it feedback, tell it to save the feedback, and it will remember. Cuz as we build this automation, this agentic workflow, it'll

[03:45] be able to do it on its own. If you do those three things, the AI stops giving you generic garbage and starts giving you exactly what you need. And notice how at none of those steps we actually tell it how to do the work. We begin

[03:57] with the end in mind. It's its job to figure out how to get there. The CEO of productivity tools out there, he calls it the manager of infinite minds. When you have this army of infinite minds out there accomplishing, researching,

[04:11] can sit in your cockpit and being the director, the future's going to look completely different than how you got to this point in your life. And again, I'm and it's super simple towards the end of this video. If you're a business owner

[04:25] figures and you're kind of struggling figuring out ways to actually integrate AI into your business, I created my AI implementation workbook for a lot of my friends because I wanted to walk them through how we integrate AI into every

[04:38] department in my company step-by-step. So, if you want a free copy, just find me on Instagram at Dan Martell and DM me the words AI business and I'll send it right over. There's a bunch of AI agent tools out there and they can confuse

[04:50] you. So, I'm going to tell you which one is right for you. Point number three, which tool is the right tool? If you ask AI to tell you what tools you can use to create agentic systems, it'll give you a bunch of different tools cuz

[05:03] there's a bunch and they're coming out every week. And every tool is great for them all, but that would mean you'd be a master of none. What I'm going to recommend is you go deep on one. Become an expert in that. I'm going to give you

[05:16] the answer to which one is right for you based on who you are, but once you know that, I want you to go all in. So, here's how we pick the right tool. If research, content, and general tasks done, use Manas AI because it is the

[05:31] best all around agent for getting real work done at the filming of this video. If you're a creative who writes, designs, or works with ideas, use Claude Co-work. They call it Co-work because it actually runs on your computer. It can

[05:45] folders, it can open up tabs on your browser, and essentially can accomplish involved. Now, if you're a developer or someone who writes code, use Claude code. It can fix bugs, it can add test files, it can do work in parallel to

[06:01] your code base while you focus on other harder problems or talk to customers or test what it's built. It is the tool that all the developers at the other AI companies use to build that AI. It's kind of wild. And if you want a

[06:14] completely automated, full personal assistant that runs on your computer and you're a little nerdy, use Open Claw. The reason why is it runs like a person. has memory. I think it gives you a glimpse of what the future will look

[06:30] like at scale. But it is a little technical to set up and truthfully risky. [music] So, be careful. I saw an example where the bot bought a $3,000 course for itself so it could be a better agent with no permission. The

[06:44] learn them all right now. I know you're excited and you're like, "Oh my gosh, use the one, the more you ask it to teach you how to use more features, the better you'll get at directing AI, which is the skill we need to develop. Let me

[06:58] actually show you exactly how you can use AI agents cuz I'm going to go through a real task with Manis right now. I'm going to make it easy for you to understand so you can follow along and see exactly how this works. Point

[07:10] number four, Manis in action. So, for example, if I'm the founder of a do some research on my competitor so I can see what they're doing and figure offer, I can get Manis to do that. So, just go to Manis and type something like

[07:25] this. Research the top three niches digital agency in Canada finding their doing well. Then create a simple one-page website that summarizes the research. It's finding those three companies. It's building a task list so

[07:40] they can follow. It's It's writing the code of the new website for the output. keyboard. [music] And I didn't give it any expert detailed instructions. Even about anything could have wrote what I wrote and they would have got an expert

[07:54] level output. Most people would have took a week plus and they probably would have confused it. This did it automatically in less than 10 minutes. And my favorite part is within Manifold,

[08:06] there's some things that are missing or maybe I want some changes. So, I'm going to tell it, "Now add a section to each one that includes client testimonials." Enter. Watch. It's going. It's literally updating itself. And just like that, I

[08:22] pulled from each of those companies, did the research and it updated the website in minutes. So, now I'm going to ask it to send it to Sam Gadget, who's my creative director, in Slack and ask him to provide feedback. And then also email

[08:35] in the research to Joel Harrison. Now I go check Slack. It posted it as me and it has a little icon that says Manifold, so they know that it was from Manifold. I can now [music] get feedback from my team and apply it to that website. So, I

[08:48] can tell it to monitor that Slack thread and see if there's a reply every 15 tell Manifold. And if it gets feedback, it's going to either prompt me or I can just say take any feedback and automatically apply it and it'll apply

[09:01] It did the research. It created the website. It shared it with my team. It took my feedback, updated it all within minutes. And what's cool is you can kick off a bunch of those in parallel and they'll all run while you're doing

[09:16] something else. So, now you got to understand, I went from an idea to a prompt to sending it out to people I work with, to getting the feedback, to updating the output. And I can even say now email or post on Slack to everybody

[09:28] that's been involved letting them know the changes made. I do this all the time. It's amazeballs. It's crazy. The fact that without knowing how to create automation, you just created automation. My favorite part is mobile. So, on my

[09:42] it and I can tell it what I want to do. Spins up, runs it, connects the stuff, pulls everything. Now, here's a pro tip. I challenge you to stay in the tool. See, we've been taught because of the chat level that we need to chat and it

[09:57] response and then we go do something with it. Post it in Slack, send an email, create a Google Doc. Don't do it. I dare you not to do it. You'll be tempted to take action. Don't. That way you get better and the AI gets better

[10:11] because it's doing more for you and it's learning how you like to work. So, now that you've seen how it works, this is the question. What are you going to do about it? What are you going to change? AI is the opportunity you've been

[10:23] why wasn't I early to the internet? Oh my gosh, I wish I would have came out with that mobile app idea. This is now. In the next 5 years is the AI gold rush and there will be more millionaires created in the next 5 years than the

[10:37] history of the internet, just like the dot com boom. This is bigger. This isn't about being an AI expert. The cool part is AI is the AI expert. It's about being willing to learn. It's about telling yourself, I need to just sit down and do

[10:52] and I know this still confuses me, but I'm willing to read, I'm willing to research, and I'm willing to empower the tool to help me in my life. And the best part is as you just saw how easy it was to do something that seems complicated

[11:06] pretending not to know. I know it's overwhelming, but if you just take action, I promise you it will click. So, here's what I want you to do right now. Pick one tool based on what I said

[11:18] earlier. Number two, pick one task that you're doing every week that takes you a lot of time. Three, have the agent do it for you today. Sit down with it, play with it. If you do that, you will will to change the mindset and the belief and

[11:33] understand how easy this is. And you will get ahead of 99.9% of everybody else using AI to do chat. And remember, if you want my free copy of my AI implementation playbook that we use in all my companies, just DM me the word AI

[11:46] business on Instagram and I'll send it right over. And if you want to learn what are the best AI tools to make money in 2026, click here and I'll see you on in 2026, click here and I'll see you on the other side.