Transcription
For 100 years, marking that one thing, you execute. Now, the job's splitting into two. One marketer keeps executing, the other commands AI agents that outproduce a whole team. I'll show you the split, where it's headed, and how to land on the right side of it.
And this isn't something for the distant future. It's already happening. In 2024, the number of day-to-day worker decisions being made on their own by AI agents was zero. By 2028, Gartner projects that it jumps to 15%. Zero to 15 in just 4 years. That's not a prediction about robots. That's software running your campaigns and your content starting to make calls on its own.
And the companies that saw this early on are already pulling ahead. We track the ones that invested in building for this. In 2024, their return was -28%. Honestly, it looked like a bad bet because the market wasn't ready. When you're later, the same strategy returned positive 144%. That's how fast this is moving, and the money is following it.
In a survey of over 9,000 marketers, 61% of B2B marketers are increasing their spend in 2026. But here's what matters. They're not spending that money across more channels. They're concentrating it into fewer. You can only do that if the people running those channels suddenly have way more leverage per person than they used to.
So, where's the leverage coming from? It comes from one change. The marketing job is quietly splitting into two completely different roles, which I call the operator versus commander. There are two kinds of marketers now, and you want to be very clear about which one you are.
The first is the operator. The operator executes campaigns. They write the copy, run the ads, manage the calendar. Their output is the task itself. You pull them out of the workflow and the work stops. Most of the best markers I've ever worked with spent years as operators. I did too, early on. It's a real skill.
The second role is the commander. The commander doesn't do the work. They direct agents that do the work. They set the strategy, point the agents at it, and make the judgment calls the machine can't. Their output isn't a task, it's a system that keeps producing whether they're in the room or not.
Here's what a single Tuesday looks like for each one. As a operator, you've got a campaign to move. You spend the morning writing five ad variations, you spend the afternoon pulling the performance data by hand, you end the day sending a status update. Eight hours and you've moved one campaign one step.
As a commander, you've briefed your agents Monday night. Tuesday morning, you open your laptop and five variations are written and a report already built. You spend three hours reviewing and deciding what ships. Five times the output in less than half the day.
And this isn't theory. Real companies run this way right now. Quanto manages over a half a million businesses across four markets in Europe. They built localization agent that adapts their content for every market. Localization time dropped up to 70%. So, check this out. Their content managers didn't lose their job. They stopped translating their text and moved up to content strategy. The agent took the grunt work, the humans took the thinking.
IBM did the same thing internally with an HR agent that now handles 94% of their basic HR questions. What's left for the humans is judgment. So, that's the split, and building these systems, turning team of operators into commanders, is exactly what we're doing right now for the brands that we work with at my agency NP Digital. If you want us to just do it for you, go to npdigital.com.
Now, you might be sitting there thinking, "This still feels a little abstract." What is an agent actually doing on a typical Tuesday afternoon? To understand this better, let me show you what agent-driven marketing actually looks like.
First, let's clear up one thing because people mix these up constantly. What's the difference between AI agents and AI tools? A tool responds to a prompt. You ask, it answers, you ask again. That's your ChatGPT tab today. An agent takes a goal, breaks it into steps, and works across your systems without you holding its hand at every stage. That difference is everything.
Here's what it looks like dropped into a real marketing team. A campaign gets approved in one market. Before anyone touches it, an agent has already generated the localized versions for five other markets, adjusted the tone for each region, and pinged the manager that the assets are ready. The first 80% happen with no human in the loop.
Meanwhile, another agent is reviewing every support ticket and reviews it as it comes in, flagging which accounts are about to churn. The thing that used to be a two-week project once a quarter is now a feed that refreshes every single morning, if not every single minute. Everything I just described is agents working for you, the ones Commander runs.
But, here's what almost nobody's accounting for. Your customers have agents now, too, and theirs are pointed right at you. And this is extraordinarily because AI platforms like ChatGPT are driving less than 1% of total website traffic. I know that sounds tiny, almost ignorable. Except, the same AI platforms drive 9.7% of B2B revenue and 11.6% of B2C revenue. Less than 1% of the traffic, almost 10% of the revenue.
How? Because by the time that buyer lands on your page, an AI already did the research for them. It compared you to your competitors. It pre-qualified you. The customer shows up halfway to a decision you never even saw being made. That's where this is going. The buying is starting to happen inside AI before anyone ever clicks.
So, now there are an agent on both sides of the table. The ones you command and the ones deciding whether your buyer ever hears your name. And only one kind of marketer is built to win that. The commander.
And if that's true, then the skill that wins from here isn't the one most people are chasing. So, here's what that means for you. Almost every conversation about how AI and marketing right now is about tools. Which one's best, how to prompt it, and I get it. That's what's visible. But, you can bolt 10 new AI tools onto your workflow tomorrow and still be a pure operator. More tools doesn't make you a commander. A different job does.
The real shift is going from doing the work to directing it. The operator wakes up asking, "What do I need to write today?" The commander asks, "What systems do I need to set up so this gets done consistently better than if I'd done it by hand?" You stop being the worker and you start being the architect.
And I'll be honest, that's uncomfortable for a lot of really good markers because they built their whole careers on being excellent at the execution. Stepping back from it feels like giving something up. It isn't. It's a promotion.
And here's the skill that actually matters in this new world. It's not prompting, it's taste. When an agent hands you 10 pieces of content, the entire game is being able to look at them and instantly know which one is right. Which one is on brand? Which one our customer will actually trust? That judgment can't be automated, and it gets more valuable the more those agents produce because someone has to decide what's worth using.
The commanders who win, they don't just switch agents on and walk away. They build the checkpoints. They know where the human has to stay in the loop and where it's safe to let the machine run. IBM's own team will tell you the more you automate, the more your judgment and your guardrails become the thing that separates you from the company that gets burned the first time an agent goes off the rails.
And this is already showing up in the numbers. Across 22 companies we tracked, leads coming from AI-powered search went up 3.1% of total leads to 7.4% in a single year. It doubled. The commanders built systems to catch that channel. The operators are still staring at old dashboards wondering why their traffic isn't turning into pipeline.
So, what do you actually do with this? You don't need to go and automate your whole department this week. Start smaller than that. Look at your own calendar for your next 5 days and be honest about how much of it is pure execution. Writing by hand, pulling reports by hand. If that's most your week, you're operating like an operator in a market that's about to pay for commanders.
Then pick one piece of that and practice handing it off and directing it instead of doing it. And change your scoreboard. The old one was rankings and click-through rate. The new one is whether you even show up when the AI answers a buyer's question. You can actually check this out by going to ubersuggest.com. And there's a free AI visibility report that shows which of your pages are getting pulled into AI answers and which ones are completely invisible. That's the number I'd be watching now.
And when people run that report, it tells them something uncomfortable. The AI can't see them at all, which is what this all comes down to. Because commanding your own agents is only half the game. The other half is whether your customer's AI agents trust you and have to recommend you at all. That's who you're selling to now, not just the buyer. You're also selling to their agent as well. How you earn that trust is the next video. Go and watch it.