AI Agents for Beginners: Market Growth and My 12-Agent OpenClaw Setup
Prefer to watch? Watch on YouTube · Subscribe for new builds · Read the transcript
In this video I explain why AI agents are getting so much attention, why I think we're still early, and what running 12 agents on a Mac Mini looks like day to day. It's for people who keep hearing about agents but haven't built anything yet. I wasn't technical either when I started.
I recorded this in March 2026. Tools change fast, so check the current docs before you set anything up.
What I cover in this video
- 0:00 Why the agentic AI market is projected to grow from about $9 billion to $139 billion
- 0:46 What an AI agent is and how it's different from a chatbot like ChatGPT
- 3:02 What NVIDIA and Zendesk are doing with agents
- 3:48 My 12-agent setup on a Mac Mini and what a typical morning looks like
- 6:05 My background and why I think anyone can do this
- 9:09 Is it too late to get into AI agents?
- 9:54 Three things I recommend doing right now
- 12:55 Closing thoughts on what you're going to build
Chatbot vs. AI agent
The simplest way I can put it: a chatbot talks, and an agent actually does the work. With ChatGPT you type something, get a response, and the conversation is over. With an AI agent you hand it a task and it goes and does it. It can send emails, research competitors, write content, monitor systems in your business, and schedule meetings, all without you sitting there going back and forth with it.
That's why businesses are paying attention. They can hand whole workflows to agents that run 24/7 for a fraction of the cost.
The market numbers
A market analysis out the week I recorded this put the global agentic AI sector at about $9 billion now, heading to $139 billion over roughly the next 8 years. That's about a 15x increase with a compound annual growth rate around 40%. I compare it to when smartphones first came out and the people building apps early did really well, or when social media started and early creators built big audiences.
Two other data points I mention:
- NVIDIA says 64% of organizations are already deploying AI agents in production. They're past the experimenting stage.
- Zendesk spent $200 million acquiring Forethought because they believe AI agents will handle most customer service interactions.
My 12-agent setup on a Mac Mini
My setup is much smaller than that. It runs on the Mac Mini behind me, which cost maybe five or six hundred dollars to get going. I run 12 agents in production every day across my content pipeline, community management, security audits, and most of my social media. I use OpenClaw with Claude as the model.
Here's what a typical morning looks like:
- 5:00 a.m. My intel agent wakes up and gathers what's happening on Reddit, X, and YouTube. We also use TubeLab to check outlier scores for thumbnails.
- 6:00 a.m. A writing agent takes that intel and creates copy and outlines, including four scripts a day. At the same time another agent drafts five X posts.
- 6:30 a.m. A thumbnail agent creates a set of thumbnail options based on what's working and drops what isn't.
- 7:00 a.m. One agent writes a blog post and another scouts for podcast guests.
By the time I get my coffee, I have outlines, scripts ready to record, thumbnails to review, and posts scheduled. Nothing goes to the web automatically. The agents get me about 90% of the way there, and then I check everything and look at what could be better. I used to do all of this by hand, so it saves me a lot of time.
Three things I recommend doing now
1. Pick a framework and start building
It could be OpenClaw, Claude Code, or ChatGPT. Build something and break something. For me it's OpenClaw. It's an orchestration layer that sits on top of large language models, it's open source and free, and I find it a lot less intimidating once it's running. It connects to Telegram, WhatsApp, iMessage, and Discord, so I can talk to my agents from anywhere. I got back from church and was able to launch clips, check on thumbnails, and scrape Reddit posts from my phone.
I also like the soul markdown file, which gives your agent more personality so it feels more like talking to a person. Because I use Claude, every time the model improves, my whole system gets better too.
2. Automate something real
Don't build another toy chatbot. Pick something in your business or personal life that takes up your time and build a system around it. For me that was the content pipeline, because I was spending hours a day on research, scripts, outlines, thumbnails, and social posts.
3. Join a community
Learning from other people's mistakes saves you a lot of time, kind of like reading a book instead of figuring it all out yourself over years. Some days I'm pulling my hair out trying to figure out why an agent broke or why a cron job failed at 3:00 a.m. Having people around who are dealing with the same problems helps you debug faster and keeps you accountable.
Why I think anyone can do this
A year ago I couldn't write a line of code or use a terminal. I bounced from job to job in my twenties. In 2020 our house burned down, I lost my job, and we were starting a family, so I went all in on learning entrepreneurship. We built a reselling business that now does over seven figures a year, and I took the same approach to AI: start and figure it out as you go. I'm not a developer and I don't have a computer science degree. If I can run 12 agents on my desk, you can build something too.
Key takeaways
- A chatbot answers questions. An agent takes a task and does the work without you going back and forth with it.
- The agentic AI market is projected to go from about $9 billion to $139 billion in roughly 8 years, so we're still early.
- You can run a useful agent team on a single Mac Mini. Mine handles research, scripts, X posts, thumbnails, blog drafts, and guest scouting every morning.
- Keep a human review step. My agents get me 90% there and nothing publishes automatically.
- Start with one framework, automate a real task you already do, and find people to build with.
FAQ
What is the difference between an AI agent and a chatbot?
A chatbot like ChatGPT responds to what you type and then the conversation ends. An AI agent takes a task and goes and does the work, like sending emails, researching competitors, writing content, or scheduling meetings, without you going back and forth with it.
How big is the AI agent market expected to get?
The analysis I reference puts the agentic AI market at about $9 billion now, growing to $139 billion over roughly the next 8 years. That's about a 15x increase with a compound annual growth rate around 40%.
Can you run multiple AI agents on a Mac Mini?
Yes. I run 12 agents in production on a Mac Mini that cost maybe five or six hundred dollars to get going. They handle my content pipeline, community management, security audits, and most of my social media.
Is it too late to start building AI agents?
No. With the market around $9 billion and projected to reach $139 billion, I think we're still in the first inning. The best move is to pick a framework like OpenClaw or Claude Code and automate something real.
Chapters
Transcript
Read the full transcript
0:00Everyone and their grandma is talking about AI agents and how they all work and you've heard it in the news and in this video I'm going to talk about why that is and why we are still early. And I'm going to talk about the market cap. We are projected for a $139 billion market cap. That's how big the AI agentic workflow is projected to be here in the next 8 years. Right now it's sitting at about 9 billion. So we're talking about a 15x increase in less than a decade and a compounded annual growth rate of 40% So to put that into perspective for you, we are in that smartphone market area when smartphones just came out. We're in the stage where people who built apps on top of smartphones, they became millionaires. We're in that era of when social media
0:46would just came out and the people who started creating content early built empires. We are at that same inflection point right now with AI agents and most people are still sitting on the sidelines. If you are one of those, it's okay. I was those type of person this time last year. So let's break it down and back up. What is exactly an AI agent and how is that different from chat GPT and talking to just another chatbot? The simplest way I can explain it is a chatbot talks, an agent does as well, but when you use like chat GPT, you type something in, it responds and you know you get a response and then it's done. The conversation's over. When you use an AI agent, you give it a task and it goes and does the work. It can send emails,
1:31it can research competitors, it can write content and monitor a lot of the systems that are going on in your business. It can also schedule meetings, it can do a lot of those workflows that we do every single day. But it can do all of this without you versus you sitting there going back and forth with it. And that's why the market is exploding because businesses are realizing they can replace entire workflows with agents that run 24/7 for pennies on the dollar. And I'm going to tell you why, you know, what I'm actually seeing in this space and what I've learned over these last few months. This isn't just me guessing hypothetical stuff. This is actually stuff that's happening right now. So, this week a new market analysis just came out showing the global agentic AI sector is on track for that $139 billion market cap and the
2:17growth rate is not stopping. This isn't just like a projection. This is companies actually working with AI agents and deploying them into production right now. Real quick, before we keep going, if you're watching this and you want to actually build with some of these tools, not just watch videos about them, then you're going to want to check out our community down below, Shipping School. We have a full Claude code course, a full open Claude course, and four live bootcamps every single week where we actually help you get set up from scratch. Like actually set this thing up, not just watch a tutorial and figure it out by yourself. And we also provide one-on-one coaching, so you could book a call with me. We could share screens and I can help you get Claude code or open Claude running on your machine. That's it, no fluff. I
3:02built this community because watching YouTube's only gets you so far. We launched it just 3 days ago and we have over 55 members. You need people around you who are actually building, people who hold you accountable, and coaches who can help you when you get stuck. I'll put the link in the description down below. Get in now before the price goes up. Nvidia, pretty big company, says that 64% of organizations are already actively deploying AI agents. They're not experimenting, they're not creating tutorials, they are actually like deploying them inside of production. Zendesk, a huge company, just spent $200 million acquiring Forethought because they believe AI agents will handle most customer service interactions. So, that's not a side project, that is a
3:48$200 million gamble they are taking. The funniest part about all this is I've been doing something similar on the Mac Mini behind me that cost maybe five, six hundred dollars in total to get off the ground in my time. But running 12 agents in production every single day, my content pipeline, the community management, the security audits, basically all the social media, it's working. But let me walk you through what a typical day looks like for me and my agent crew. We have at 5:00 a.m. Ghostface wakes up and does a lot of the intel. Starts gathering intelligence from Reddit, X, YouTube. We use Tube Lab to help figure out outlier scores for thumbnails as well. Checks X for what's popping off. Pulls the data from multiple sources. And then at 6:00 a.m. the Rizza picks it up and starts working
4:34with that intel creating copy, creating outlines, creating a lot of the content. So four scripts a day at the same time, all that's done and another agent wakes up. Inspect the deck is basically crafting those five X posts. And by 6:30 a.m. You God is in charge of the thumbnail management. Starts creating a variety of thumbnails of what's working and discarding what's not working. So by 7:00 a.m. another agent is writing a blog post and another one is scouting for podcast guests. So basically all this happens right when I wake up. I get my cup of coffee. I check the work. And I will preface this by saying nothing goes to the web automatically. It gets me 90% there and I check the work. I make sure everything looks good. What
5:19could have been improved upon. So it does save me a lot of time in the long run cuz I used to have to do this before manually. But basically I have outlines ready, scripts ready to record, thumbnails ready to review and post scheduled. Basically a full pipeline loaded. And that's kind of what the AI agents look like on my end and how they work in my production setting. It's not like a full chatbot answering questions. It's a full team running the back end of my business. So here's the thing that really gets me excited about this space and I just want to be honest with you about showing the journey on my end and what we're actually doing. A year ago, I couldn't write a line of code. I didn't know what to do in the side of a terminal. I didn't know how to install Claude code. I was from the outside looking in watching other people build cool stuff. And then, you know, I got me
6:05thinking, I need to get my hands dirty. I just need to start. You know, I come from a background from working from job to job to job, you know, from ages 20 to 30. And over the last 6 years since 2020 and, you know, a lot of stuff happened in 2020. Our house burned down. I lost my job. We were starting a family and the pressure was on. So, I just really needed to dive in deep to learn entrepreneurship. We learned about our reselling business that produces now over seven figures in revenue every single year. And I kind of applied that same logic to this AI space. I just need to start. I'll figure it out as I go. And I just share that with you because I want you to get a little context of who I am and my backstory. Maybe relates to you and give you some encouragement that you could do this as well. So, like I'm not a developer. I didn't get a computer
6:51scientist degree or anything like that. I'm just a guy who decided to figure this out and tried to learn new skills as fast as possible. So, if I could build a 12 agent system running on my desk, pretty much anyone can. That's not like a motivational talk or anything like that. It's just an encouragement piece from me to you that you could do this. So, the tools exist now to build something so robust and they're getting better every single day. Open claw, the thing is with it is it's open source. So, it's free. Claude, the AI model that I use is also getting better every single month. So, at the same time that I'm using these frontier models in my system, every time that they improve in the context window or the token usage gets more optimized, I get an upgrade in my business. And also real quick, in the link down below, shippingschool.com,
7:38there is a use cases tab where it actually updates every single day on people who are actually building with open claw. You can go through here. There's automations, content creation, developer, e-commerce, education, finance, real use cases. So, you can actually click on here and view the original source. It can be helpful for people to maybe get some ideas on what other people are building. And I thought it would be relevant to this video to show you that we have this updating every single day. You could bookmark it and you can see people actually building real businesses making real money with open claw. So, like I said, that's linked down below. Check it out, bookmark it, and you can see what people are building. So, the community around this stuff is also growing pretty fast. If you've ever
8:23been in the open claw Discord, it's popping off the chain in there. So, let me give you some numbers about my own journey. Shipping School, our community, we just launched 6 days ago, just hit 70 members producing over $2,300 of monthly recurring revenue. And the YouTube channel, which was at a thousand subscribers 6 days ago, is it just about to hit 4,000 subscribers, and we just got monetized. I know it doesn't sound like a ton of money or a ton of people. I'm just saying that none of this was possible if I hadn't have leveraged AI agents to help me do the work. That's a lot of progress in a short amount of time. So, I don't I know it doesn't sound impressive, but at the same time, it's just been 6 days. So, basically, when people ask me, like, is it too late to get into the AI agent game? I laugh
9:09because I love that question. You are still so early. Then the market cap right now is $9 billion for AI agent and gen tech space, but we're headed to 139 billion in the next 8 years. So, we're sitting in the first inning of a nine-inning stretch. We're in that first mile of the marathon. We are so early. So, the people who are getting into this right now are the ones who are actually going to look like geniuses in just a few months and years. So, what should you like actually do about this? Because I've been on the sidelines years ago thinking what should I do? Should I do this, do that, do this? Analysis paralysis. But it wasn't until I started doing things and actually taking action is when things changed. So I'm going to give you the three things that I recommend doing based on my experience.
9:54I would say first thing is picking a framework and start building. Whether it's open claw, cloud code, chat GPT, just build something, break something. I think for me it's open claw. It's like an orchestration layer that works on top of a lot of these large language models, a lot less intimidating and once you get it going, it's 100% worth it. What I like is it connects to Telegram, WhatsApp, iMessage, Discord. You're able to communicate with it from far away. I would just came back from church and I was able to communicate with my AI agent here to launch clips, to see how thumbnails were coming along, to scrape Reddit posts, things like that. And that's what I really like of this technology is that we can communicate it from anywhere. Another thing is the soul markdown file. This can give your chatbot or LLM a little more
10:40personality. It gives you some, you know, it makes you feel like you're talking to a person, I guess. So whenever you whatever you choose to do, I think that just starting and stop watching everyone else do stuff with AI, I think that's the most important part. Just do something, start before you're ready, punch failure in the face and just go. The second thing I could recommend is automate something real. Like don't just build a toy project, don't just make another chatbot that everyone else is making. Pick something in your actual business or in your personal life that takes you time and try to automate a system around it. The first thing for me was automating the content pipeline because I was spending hours a day doing research, writing scripts, outlines, thumbnails, and posting to social media. Basically my
11:25agents handle all that heavy lifting while I focus on recording and helping people in our school community. The third thing I can recommend is joining a community because I see this as like reading books and trying to learn new skills. What could take you 5 years to learn, you could have just read in a book that takes you 3 months to apply. And that's how I see with community. We get to learn from people's failures, what people are doing, building cool projects, connect with potential founders. That's the cool part about being in a community is we're all going through the same thing together. You also get held accountable. So, there are days where I am pulling my hair out and trying to figure out why my agent isn't working and why a cron job failed at 3:00 a.m. I think it's happening to most people right now, but at least we get to do it together. And having people around
12:10you who were going through similar things can help you debug. They could share with what's working and what's not, and that does change everything. It helps you get some more of your time back rather than figuring it out on your own. And that's what we built Shipping School for exactly people like that that want to get their hands dirty and they want to start going. We do four live bootcamps a week and we work with you hands-on to get your items shipped and ideas built and maybe help you troubleshoot. So, basically, it's not just a course where you watch videos and forget everything. This is a community where we actually focus on building. So, back to the $139 billion talk, this is insane. That's where the market is headed and the question isn't whether AI agents are going to be a big deal and be a pump and dump just like every crypto scheme. That's already been decided
12:55about AI agents is they are being implemented now and it's getting validated as the market progresses. The question is is what are you going to build? What are you going to do about it? What problem are you going to solve? What businesses are you going to automate? Because the people who figure this out right now in 2026, while we are still early, are those people are going to be telling their story 3 years from now going they are so excited that they just started and got going. Where everyone else is going to be wondering what the heck happened. And that's exactly what we're doing here. We're building in public on the channel. I'm sharing my journey. I want to be 100% transparent with you. I'm just a guy just like you trying to get it going, trying to figure it out, and trying to work with people to help grow and get better one step at a time. So, making
13:42mistakes and learning from them is probably the best thing on how you learn the fastest. It's okay. Don't have fear fear of failure. You can do this. And the best part actually is that we can have fun doing it also. This isn't like, "Oh my, this is terrible. I hate doing this." I have not had this much fun in years since we first started our reselling business. It's fun and it feels a fulfilling to build something and to figure stuff out that maybe a little maybe a little intimidating from the start. You could do it. I'll put a link to the community down below and also on my website shippingschool.com, there's a use case button. You could see what people are building with Open Claw and how they're automating their life. That should be some encouragement for you as well. And if you guys haven't
14:27subscribed already, we're releasing four to six videos every single day. Hit the bell icon so you get notified on the next release. And we'll see you in a few hours. Be blessed.
Ready to start building with AI?
Join the Shipping Skool community free. Paid members get live bootcamp calls every week and weekly-updated courses.
Join Shipping SkoolKeep reading
Build an AI Agent Team With OpenClaw: Augmentation vs Replacement
I run a team of AI agents through OpenClaw on a Mac mini, with Atlas on Opus 4.6 coordinating. They handle trend research, scripts, newsletters, X posts and podcast outreach, and I review everything. My Mission Control dashboard, a Next.js app, tracks all of it. I use AI to boost my own work instead of replacing people.
OpenClaw Marketing Agents: A 7-Agent Stack for Solo Founders
An indie hacker named Rithik built 7 single-purpose OpenClaw agents: an orchestrator plus agents for Reddit research, writing, repurposing, tweets, trend monitoring, and engagement. He says it took his marketing from 15-20 hours a week to almost zero. My advice: start with one agent and add more over time.
OpenClaw Backup and Rebuild: Restoring 12 AI Agents From Scratch
After a power outage forced a fresh OpenClaw install on my Mac Mini, I rebuilt all 12 agents in about 2 hours instead of the original 35 days. Backups in Neon, Supermemory, and local memory files did most of the work, and I moved every API key into a locked environment file.
