AI Is Getting Better at Marketing.
That Might Make Marketers More Important.

By James Arnold
Rooster Chief Digital Officer

I have spent a lot of time talking about AI over the last couple of years. Probably too much time, depending on who you ask.

But for the first episode of AMiDA 2026, I wanted to get beyond the conversation about whether AI is going to change marketing. That seems pretty well settled at this point. Instead, I wanted to talk to smart marketers who are actually using these tools and ask a more practical set of questions. What has changed? What are they doing differently? Where is AI genuinely making their work better? Where are we overdoing it? And, maybe most importantly, what do marketers need to understand now to be good at their jobs a few years from now?

I talked with Megan Ratcliff, founding partner at Clarity and Motion; David Mainiero, chief AI officer and general counsel at AI Digital; Rooster's Jessica Robinson; and Ryan Green, founder and chief media officer at The Last Splash. They approach AI from very different perspectives, but I was struck by how often their answers came back to essentially the same idea.

AI can do an incredible amount of work. But knowing what good work looks like is becoming more important, not less.

We need to get past treating AI like a shortcut

Megan Ratcliff made a distinction that I think is going to become increasingly important. Knowing how to use AI tools isn't the same thing as knowing how to work with AI.

“AI literacy is not the same thing as AI fluency,” she told me.

That's a pretty important distinction because I suspect a lot of us, myself included, have spent the past couple of years building literacy. We've learned how to prompt. We've tried ChatGPT, Claude, Gemini and a growing list of specialized tools. We've figured out that one model might be particularly good at one task and another better somewhere else. We've watched the tools get substantially better in a remarkably short period of time.

But fluency is different.

Megan described one of the biggest changes she's seeing as the move toward persistent context. Instead of opening an AI tool, having a conversation and essentially starting over the next time, organizations can begin creating a shared layer of knowledge about who they are, what they do, how they talk about themselves and even what they don't say about themselves. That context can then become available across the organization.

That's when AI starts becoming less of a tool you occasionally use and more a part of how the business operates.

As Megan put it, persistent context allows companies “to stop using AI as like a quick little shortcut,” because it can instead become operational across the business.

Interestingly, the first step toward that isn't especially technical. Her advice was basically to gather your team and write down what you want people, and ultimately the AI, to know about your organization. What do we believe? How do we describe ourselves? How do we talk about our services? What does our brand sound like?

The technology comes after you figure out the answers.

That feels like an important lesson for marketers. We tend to get excited about the tool. The harder work is still deciding what we want the tool to know and what we want it to do.

The advantage isn't the AI. It's what you bring to it.

David Mainiero described AI as “a leverage tool, not a replacement for thinking.”

I love that framing.

There is something seductive about giving an AI a problem and watching it return a 20-page research report, a strategy, a piece of code or a polished-looking presentation a few minutes later. It can be difficult not to equate the speed and completeness of the response with quality.

But David's point was that marketers still know things the AI doesn't. We understand our customers. We understand our clients. We know the history behind a decision, the personalities involved, the vendor relationships, the weird thing that happened three years ago that everyone in the room still remembers. AI may have access to more information than any of us, but information and understanding aren't quite the same thing.

“You're the orchestrator,” he said.

That might be one of my favorite descriptions of what a marketer's relationship with AI should become.

Instead of asking AI for an answer and accepting it, ask it to challenge the answer. Pressure test the strategy. Play the skeptical client. Find the holes in an argument before somebody else finds them in a meeting. Give it your expertise and then use its capabilities to push your thinking farther.

David put the downside in slightly more colorful terms: “If you're using it to avoid the hard work of strategy and judgment, and using your taste as a filter, you're then using it to make yourself replaceable. And I'd rather see everyone use it to make yourself dangerous.”

There is a pretty good career strategy buried in there.

Start with the thing you already know

Ryan Green made a related point that I hadn't thought about quite this way before.

If you're trying to figure out where AI fits into your work, start with the thing you're already good at.

Ryan spent years in paid search and programmatic media. Today, he uses AI on every paid search campaign he manages. But that works particularly well because he has roughly 15 years of experience telling him when the AI is right, when it's mediocre and when it's completely full of it.

“You will know what is [bad] and what's not,” he said. “You're going to know where this is quality work and where it's not.”

That makes a tremendous amount of sense.

The temptation with AI is almost the opposite. We immediately want to use it to do the things we don't know how to do. I can't code, so write me an application. I can't design, so make me an ad. I don't know much about SEO, so give me an SEO strategy.

And you can do that. That's part of what makes these tools so remarkable.

The problem is that if you don't know the subject, you also don't necessarily know where the bar is.

Ryan gave a great example from creative work. AI can make a video. But an experienced art director knows the difference between a C-minus AI-generated video and something worthy of running as a television commercial. The tool didn't eliminate the expertise. The expertise is what allowed someone to get considerably more from the tool.

Ryan has a simple way of thinking about it: figure out where AI produces A work, where it produces B work and where it produces C work. Then understand where your own expertise is needed to get the result to an A.

I think that's considerably more useful than asking whether AI can “do” a particular marketing job.

Of course it can do some of it. The better question is how well.

And please, let it do the plumbing

There is another side to this that shouldn't get lost in all the talk about strategy and human judgment.

Some marketing work is just tedious.

Ryan calls it “the plumbing,” and his example immediately took me back to plenty of reporting exercises I've experienced over the years. Early in his career, he would spend days copying campaign numbers into PowerPoint reports. Today, he can automate much of that work and spend his time on the recommendations and insights where his experience actually matters.

“I can create 10 reports in the time that it would have taken me to do one 10 years ago,” he said.

Good.

I'm not particularly nostalgic for copying numbers between spreadsheets and PowerPoint slides. I doubt many marketers are.

If AI can take away some of that work and give people more time to think, analyze, talk to customers, develop ideas or make the work better, that seems like a pretty unambiguous win.

The danger comes when we confuse eliminating tedious work with eliminating the thinking that should happen around it.

Our audience isn't entirely human anymore

Jessica Robinson brought another wrinkle into the conversation that I think marketers, and particularly content marketers, need to get their heads around quickly.

We aren't only creating content for people anymore.

Obviously algorithms have influenced what people see online for years. SEO itself is largely an exercise in making useful content understandable and discoverable by machines. But AI changes that relationship because increasingly the machine isn't simply helping someone find the answer. It is providing the answer.

Jessica described it as “SEO but extra.”

That sounds funny, but it's a pretty good description of the challenge.

If someone asks an AI about your company, product or category, what does it say? Does it understand how your product is different? Does it describe your brand accurately? Does it pull information from sources you trust? And if the answer is wrong, what can you do about it?

That creates some very practical changes for content marketers. Jessica talked about using more natural-language questions and answers in content because they better reflect how people interact with AI. She also emphasized the growing importance of third-party validation and earned media. An article about your product from a credible outside source may influence how AI understands your brand differently than another post on your own website.

And measurement gets more complicated too. Someone may get the information they need directly from an AI answer without ever clicking through to your website.

So are we measuring clicks? Mentions? References? Accuracy? Share of AI answers?

The answer right now may be some combination of all of them. It is changing quickly enough that I don't think anyone should pretend we've solved this yet.

Maybe this is good news for marketers

I understand why AI makes people nervous. It can do things today that I wouldn't have believed possible a few years ago, and there is no reason to think the pace of improvement is going to slow down just because we'd like a little time to catch up.

But after these conversations, I'm more optimistic about what this means for marketers.

Not because AI won't replace tasks. It absolutely will. It already has.

I'm optimistic because so much of what these tools seem to do best is give capable people more leverage.

The experienced paid search marketer can manage more. The strategist can pressure test an idea before presenting it. The account person can walk into a conversation with years of institutional knowledge immediately available. The content marketer can better understand how a brand is being represented by AI. The generalist can suddenly execute work that previously required several different specialists to even get started.

The common denominator is still the person using the tool.

You need enough expertise to know where the bar is. Enough curiosity to experiment. Enough judgment to reject a bad answer. Enough understanding of your customer to add the context AI doesn't have. And enough fluency to stop treating these incredibly capable systems like an easy button.

That's a different way of working.

It might also be a considerably better one.

And we're still very early.

AI Is Getting Better at Marketing. That Might Make Marketers More Important.