Hi:
Those of you who know me know that I run an advertising service business.
It has never been easier to deliver this service than it is today. I used to rely on contractors for everything. Now I'm able to take contractor work and multiply it by 5-10x (literally), doing the rest with AI programs.
One ad done by 3 people (UGC creator/videographer, editor, script writer) is now 10 ads done only by the editor. UGC creator is now an AI avatar, or just our old UGC creator with a new AI voiceover.
One media buyer (me) can now run dozens of accounts with no additional help.
The client and I come up with the grand strategy. Micro strategy comes from ChatGPT. Tactics are the same as we've always used.
Analysis is made so much easier and quicker with large language models operating directly in Google Sheets.
And all of this has been made available by dozens of SaaS companies from all over the world, who have been in a mad rush to improve their product before the AI tsunami completely wipes them out.
But this is just the tip of the iceberg.
Agentic AI Is Here... And We Have No Phuckin Clue What's Next
ChatGPT came out in December 2022 and shocked the world. It had the power to take a task, break it down into pretty accurate subtasks, and write passable (with enough iterations) code to complete many of those tasks.
This was when people realized that, with enough data and computing power, whatever technology was making ChatGPT work miracles would eventually be able to do damn near any job that can be done with a brain and some software, and it was only a matter of time before we figured out how to do it.
What Is Agentic AI?
Agentic AI is AI with agency. That is, AI with the power to make intelligent decisions based on answers generated by a model, like an LLM (ChatGPT, Grok) or a locally trained machine learning model.
You're used to having to ask ChatGPT for everything you want, and it would tell you what to do, what to say, or what code to run. That's powerful enough. But an agent has the ability to DO the thing.
You're using ChatGPT to write emails for you. An agent would respond to a trigger (an email reply), formulate an email, proofread it, then send it. This is currently being used in customer service and is a feature of famous SaaS offerings such as Gorgias.
You're using Veo3 to generate a video to use in an ad. An agent would listen to signals of ad performance (say, an ad with a high CTR), download it, then generate 5 of them just like it, then upload them to Facebook with a .csv file containing all the same ad parameters, sending it into the approval queue.
You're using Claude to write scripts for automations. An agent would write the code AND run it in a production environment (hopefully after testing it thoroughly!)
All of these things are doable right now. It will just take a little bit of time for everyone to start doing them.
That Sounds Cool I Want To Make Them
There's lots of stuff on the Internet now about making agents.
There's the Huggingface course: Welcome to the AI Agents Course - Hugging Face Agents Course which teaches you about smolagents.
Does code scare you? There's always n8n.io - a powerful workflow automation tool which has a drag and drop no-code builder.
UiPath also offers an agent builder and orchestrator, which allows you to use multiple agents in tandem to get workflows done. UiPath automation platform: drive AI transformation with agentic automation | UiPath
That Sounds Cool I Want To Make Money With AI Agents
Here are the ways I've found to make money with AI agents.
1) Using them directly to do things that make money.
2) Building a useful agent & selling access to it.
3) Sell shovels, that is, sell material for people who build and sell AI agents.
I currently (as of June 2025) do #1.
But here's the funny thing. Once you get to the point where you want to put your AI agent into production, you run into a very interesting problem.
AI Agents Are Not Perfect, Not Even Close
While there's been a lot of hype around AI agents, people who start off excited always end up in the trough of disillusionment.
Truth is, the AI agents I use are often wrong. I don't trust them. I can't. I still have to monitor their outputs before anything they touch ends up published or sent to someone.
Sometimes, getting them to work at all is frustrating.
The reason is because LLMs are not deterministic. They don't just "come up with the right answer". They guess their way to an answer that sounds plausible.
As a result, relying on them to figure out (1) what to do, (2) how to do it, and (3) to do it right the first time just isn't realistic.
You're probably thinking: "no worries, I'll just give them my company's data and train them"
That's a brilliant idea. We have options for improving agents, such as RL fine tuning. It will probably help significantly.
But you still won't end up with a perfect agent. And it won't be easy or cheap to do.
AI Agents Will Not Erase All Jobs
SMBs, especially those with a single owner, often hope that AI will get good enough to get rid of all, or most, of their payroll expense.
But I think anyone with this expectation will be sorely disappointed.
While agentic AI is all about giving AI the power to make decisions and execute on them, there are some decisions that will still be left to humans. For example, I would never let an AI agent turn on an ad without having some human check it first, that way it's that human that is accountable for any mistakes the AI makes.
Therefore, the single most valuable skills in the post-AI agent era will be: the ability to say "yes" or "no" with accurate judgment & the willingness to take on the consequences of that decision.
And there are going to be some decisions that an AI will be able to make better than even the best human.
But some things are simply going to require humans to check, otherwise you risk a rogue AI doing something idiotic (or with intent to sabotage)
What AI Agents Are You Using (Or Making)?
Comment below...
Those of you who know me know that I run an advertising service business.
It has never been easier to deliver this service than it is today. I used to rely on contractors for everything. Now I'm able to take contractor work and multiply it by 5-10x (literally), doing the rest with AI programs.
One ad done by 3 people (UGC creator/videographer, editor, script writer) is now 10 ads done only by the editor. UGC creator is now an AI avatar, or just our old UGC creator with a new AI voiceover.
One media buyer (me) can now run dozens of accounts with no additional help.
The client and I come up with the grand strategy. Micro strategy comes from ChatGPT. Tactics are the same as we've always used.
Analysis is made so much easier and quicker with large language models operating directly in Google Sheets.
And all of this has been made available by dozens of SaaS companies from all over the world, who have been in a mad rush to improve their product before the AI tsunami completely wipes them out.
But this is just the tip of the iceberg.
Agentic AI Is Here... And We Have No Phuckin Clue What's Next
ChatGPT came out in December 2022 and shocked the world. It had the power to take a task, break it down into pretty accurate subtasks, and write passable (with enough iterations) code to complete many of those tasks.
This was when people realized that, with enough data and computing power, whatever technology was making ChatGPT work miracles would eventually be able to do damn near any job that can be done with a brain and some software, and it was only a matter of time before we figured out how to do it.
What Is Agentic AI?
Agentic AI is AI with agency. That is, AI with the power to make intelligent decisions based on answers generated by a model, like an LLM (ChatGPT, Grok) or a locally trained machine learning model.
You're used to having to ask ChatGPT for everything you want, and it would tell you what to do, what to say, or what code to run. That's powerful enough. But an agent has the ability to DO the thing.
You're using ChatGPT to write emails for you. An agent would respond to a trigger (an email reply), formulate an email, proofread it, then send it. This is currently being used in customer service and is a feature of famous SaaS offerings such as Gorgias.
You're using Veo3 to generate a video to use in an ad. An agent would listen to signals of ad performance (say, an ad with a high CTR), download it, then generate 5 of them just like it, then upload them to Facebook with a .csv file containing all the same ad parameters, sending it into the approval queue.
You're using Claude to write scripts for automations. An agent would write the code AND run it in a production environment (hopefully after testing it thoroughly!)
All of these things are doable right now. It will just take a little bit of time for everyone to start doing them.
That Sounds Cool I Want To Make Them
There's lots of stuff on the Internet now about making agents.
There's the Huggingface course: Welcome to the AI Agents Course - Hugging Face Agents Course which teaches you about smolagents.
Does code scare you? There's always n8n.io - a powerful workflow automation tool which has a drag and drop no-code builder.
UiPath also offers an agent builder and orchestrator, which allows you to use multiple agents in tandem to get workflows done. UiPath automation platform: drive AI transformation with agentic automation | UiPath
That Sounds Cool I Want To Make Money With AI Agents
Here are the ways I've found to make money with AI agents.
1) Using them directly to do things that make money.
2) Building a useful agent & selling access to it.
3) Sell shovels, that is, sell material for people who build and sell AI agents.
I currently (as of June 2025) do #1.
But here's the funny thing. Once you get to the point where you want to put your AI agent into production, you run into a very interesting problem.
AI Agents Are Not Perfect, Not Even Close
While there's been a lot of hype around AI agents, people who start off excited always end up in the trough of disillusionment.
Truth is, the AI agents I use are often wrong. I don't trust them. I can't. I still have to monitor their outputs before anything they touch ends up published or sent to someone.
Sometimes, getting them to work at all is frustrating.
The reason is because LLMs are not deterministic. They don't just "come up with the right answer". They guess their way to an answer that sounds plausible.
As a result, relying on them to figure out (1) what to do, (2) how to do it, and (3) to do it right the first time just isn't realistic.
You're probably thinking: "no worries, I'll just give them my company's data and train them"
That's a brilliant idea. We have options for improving agents, such as RL fine tuning. It will probably help significantly.
But you still won't end up with a perfect agent. And it won't be easy or cheap to do.
AI Agents Will Not Erase All Jobs
SMBs, especially those with a single owner, often hope that AI will get good enough to get rid of all, or most, of their payroll expense.
But I think anyone with this expectation will be sorely disappointed.
While agentic AI is all about giving AI the power to make decisions and execute on them, there are some decisions that will still be left to humans. For example, I would never let an AI agent turn on an ad without having some human check it first, that way it's that human that is accountable for any mistakes the AI makes.
Therefore, the single most valuable skills in the post-AI agent era will be: the ability to say "yes" or "no" with accurate judgment & the willingness to take on the consequences of that decision.
And there are going to be some decisions that an AI will be able to make better than even the best human.
But some things are simply going to require humans to check, otherwise you risk a rogue AI doing something idiotic (or with intent to sabotage)
What AI Agents Are You Using (Or Making)?
Comment below...
Dislike ads? Become a Fastlane member:
Subscribe today and surround yourself with winners and millionaire mentors, not those broke friends who only want to drink beer and play video games. :-)
Membership Required: Upgrade to Expose Nearly 1,000,000 Posts
Ready to Unleash the Millionaire Entrepreneur in You?
Become a member of the Fastlane Forum, the private community founded by best-selling author and multi-millionaire entrepreneur MJ DeMarco. Since 2007, MJ DeMarco has poured his heart and soul into the Fastlane Forum, helping entrepreneurs reclaim their time, win their financial freedom, and live their best life.
With more than 39,000 posts packed with insights, strategies, and advice, you’re not just a member—you’re stepping into MJ’s inner-circle, a place where you’ll never be left alone.
Become a member and gain immediate access to...
- Active Community: Ever join a community only to find it DEAD? Not at Fastlane! As you can see from our home page, life-changing content is posted dozens of times daily.
- Exclusive Insights: Direct access to MJ DeMarco’s daily contributions and wisdom.
- Powerful Networking Opportunities: Connect with a diverse group of successful entrepreneurs who can offer mentorship, collaboration, and opportunities.
- Proven Strategies: Learn from the best in the business, with actionable advice and strategies that can accelerate your success.
"You are the average of the five people you surround yourself with the most..."
Who are you surrounding yourself with? Surround yourself with millionaire success. Join Fastlane today!
Join Today