TL;DR

The gatekeepers are back!

For most of the last century, a person stood between every brand and every audience—journalists, editors, the front desk—and people could be persuaded. Then social media dissolved the gate, and for 15 years, paid media worked as a skeleton key that could buy any brand access to any audience.

That era is ending.

A year ago, I said AI is a new brand persona—an audience you have to understand and write for. That persona just got promoted to gatekeeper. Gartner predicts 60% of brands will run one-to-one interactions through agentic AI by 2028, and when AI agents broker the conversation, they decide who gets recommended—and there is no ad unit that reaches them.

Key Insights

Gartner predicts 60% of brands will use agentic AI for one-to-one customer interactions by 2028 and calls it “the end of channel-based marketing as we know it.”
We’ve run through gatekeepers before—journalists, editors, and the front desk controlled access to anyone who mattered for a century, and because they were people, they could be persuaded (or fed).
A year ago, AI was a new brand persona—one more audience to understand and write for. Now that persona is the gatekeeper, and it’s the first in communications history that is both the audience for your content and the referee deciding who else hears it.
The new gatekeeper doesn’t take meetings, doesn’t accept exclusives, and doesn’t sell an ad unit.
Each PESO Model® media type has exactly one job at the gate: owned states the claim, earned corroborates it, shared distributes it, and paid amplifies it.
Media relations just changed job descriptions. Third-party validation is no longer an awareness play—it’s an admission strategy, because AI engines are three times more likely to cite premium publisher content than anything a brand says about itself.

AI Agents are the New Gatekeepers

I’ve been thinking a lot about AI and where things are headed. The recent Politico article detailing the OpenAI hack is itching my brain in ways beyond marketing and communications. I’m considering the demise of our society at the hands of AI, and the thought isn’t super fun.

At the same time, my thoughts return to work (because I’m me and I can’t help it) and what it means in the short term, and here is where I keep landing.

We’ve all certainly seen a lot of change since late 2022, when ChatGPT launched to the world (or at least, the part of the world that was willing to embrace it).

I honestly don’t know how I did anything without AI in the beforetimes. I don’t even get dressed in the morning without consulting my AI, who is brutally honest about the Bears sweatshirt I adore that has seen better days.

All of this change has my brain going a thousand miles a minute. Not about what AI does for us. It’s what AI is about to do between every human we’re trying to reach.

Exactly one year ago (which is crazy that I’m this consistent), I started talking about how AI is a new brand persona for you to consider. It’s an audience you have to understand and write for, the same way you’d build a persona for a buyer.

I think I was right, especially at the time, but I was also underselling it.

Because the persona just got promoted. AI agents are becoming the gatekeepers.

AI isn’t one more audience you write for anymore—it’s the audience that decides whether any other audience ever hears from you.

And if that’s true (and the research says it is), the entire logic of how brands reach people—and how they earn trust—is about to invert.

Not evolve.

Invert.

For 15 years, trust flowed from reach. Buy enough impressions, show up in enough feeds, and familiarity did the rest.

Now it runs the other direction.

If the gatekeepers are AI agents, then the only things that matter—from a marketing and communications perspective—are what we say about ourselves and who corroborates it.

The signals you can buy count for a little less every day. The signals you can’t buy are the only ones the gatekeeper trusts.

Owned media states the claim, earned media corroborates it, shared media distributes it, and paid media amplifies it. But the gatekeeper framing sharpens the whole thing into a single, slightly uncomfortable sentence.

You will no longer be able to buy your way in.

On this week’s Spin Sucks podcast episode, we are going to discuss the new gatekeeper and how, if you can’t buy your way in with donuts or free lunches (or a crazy ad spend), what you’ll do instead. Maybe we’ll even talk about unpaid media again. It’ll never get old! 😂

We’ve Lived With Gatekeepers Before

For most of the last century, there was a gate between every brand and every audience, and a human being stood at it. Journalists. Editors. Producers. Executive assistants. The front desk. If you wanted to reach anyone who mattered to your business—a CEO, a doctor, a lawyer—you went through someone first.

Take the pharmaceutical sales rep, for example. For many decades, if a sales rep wanted to get in to see a doctor, they knew the way to do that was through the hearts (and stomachs) of the front-desk staff.

Free breakfasts and lunches. Gift cards. Sometimes, even trips. Build the relationship there first, and then, when you show up with a new drug for their doctor to consider, you’re escorted right in.

Same thing with journalists (minus the gift cards). You pitched, you built relationships, and you earned the coverage…or you didn’t get in.

That’s why media relations exists as a discipline. The entire craft was built around persuading a gatekeeper that your story deserved their audience. And because the gatekeeper was a person, the craft worked.

People can be convinced. People can be charmed. People, it turns out, really like free food.

Then social media arrived and dissolved the gate. Suddenly, every brand was a publisher with a direct line to anyone. No pitch required. No editor to convince. No front desk.

We celebrated it as democratization—and it was!—but something else happened at the same time.

Paid media became a skeleton key.

It was a different kind of key than the sales rep’s breakfast. You weren’t buying your way past a person who still had to be convinced. There was no person. You were buying the audience directly.

For roughly 15 years, any brand with a budget could buy its way in front of any audience. Targeting got sharper, platforms got hungrier for ad revenue, and access to attention became a line item. You didn’t need to earn the room. You could rent it.

An entire generation of marketers has never worked in a world where you couldn’t simply pay for reach.

That world is ending.

The New Gatekeeper Doesn’t Take Meetings

Here’s what the research says is coming, so you can back up what just ol’ Gini is saying.

Gartner predicts that 60% of brands will use agentic AI to deliver one-to-one customer interactions by 2028. They describe agents as “persistent digital concierges” and call this—their words, not mine—”the end of channel-based marketing as we know it.”

The end of channel-based marketing as we know it.

And it’s not just brands deploying agents. Consumers are, too. When your customer’s AI assistant researches the options, compares the claims, and comes back with a recommendation—or increasingly, just makes the purchase—that agent has become the gatekeeper between you and them.

Harvard Business Review’s assessment of this shift is refreshingly blunt. They said brands often can’t control these agents.

Can’t control them. Can’t charm them. Can’t take them to lunch.

Every previous gatekeeper stood between you and your audience. The reporter never bought the product. The algorithm never read the white paper.

This gatekeeper is different. It’s the persona you’ve been writing for and the referee, at the same time. It consumes everything you publish—and then decides who else gets to.

That has never happened before. The audience and the gate are the same entity.

Think about what the old gatekeepers responded to. A front desk could be won over with breakfast. A journalist could be persuaded by a great story, a sharp angle, or a relationship built over years. An algorithm could be gamed—for a while—with engagement bait and a boosted post.

But an AI agent evaluating your brand on behalf of a buyer responds to exactly two things: the claims you’ve made about yourself, and the evidence that anyone else backs them up.

Does your website say what the trade coverage says? Does the analyst mention corroborate the case study? Does the community discussion agree with the product reviews?

Every consistent answer raises the agent’s confidence in you. Every gap lowers it. I wrote last week about why this is a PESO Model® problem—AI systems reward probability, not popularity—and the gatekeeper lens explains why that’s true.

The agent isn’t ranking you.

It’s vouching for you. And it refuses to vouch for anything it can’t verify.

Why You Can’t Buy Your Way Past an AI Agent

The biggest difference affecting your 2027 budgets is that there is now no ad unit at the agent layer. When an AI assistant resolves a question—which firm to shortlist, which software to buy, which agency to call—the answer arrives before any ad is seen.

The platforms are scrambling to figure out what advertising even looks like in an agent-assisted world, and the honest answer so far is that nobody knows.

But we do know what the agents actually read, because the data is piling up. AI engines are three times more likely to cite premium publisher content than brand-owned content. They weigh third-party validation—earned coverage, analyst mentions, community discussion—most heavily precisely because brands can’t buy it.

The signals that can’t be purchased are trusted because they can’t be purchased.

That’s the inversion.

For 15 years, paid was the reliable path and earned was the nice-to-have. At the gate, it flips and earned becomes the credential, and paid becomes the megaphone you use after you’re already in the room.

To be clear—and I want to be really clear, because someone will read this as “Gini says stop advertising”—paid media still matters. I’m definitely not saying “stop advertising.” But you do want to shift how you advertise.

Paid now amplifies your best corroborated content. Paid now accelerates what’s already working.

What paid can no longer do is substitute for credibility. A brand with a great ad budget and no corroborating evidence is now invisible in exactly the conversations where buying decisions are being made.

You can’t rent the room anymore. You can only be recommended into it.

Duolingo Already Proved the Point

If you want proof that the paid skeleton key stopped working before the AI gatekeepers had even arrived, I already ran Duolingo through the PESO Model Diagnostic® to prove it.

In February 2025, brands paid roughly $8 million for 30 seconds of Super Bowl airtime. That same week, Duolingo killed its owl—and the fake funeral generated a bigger, longer-lasting spike in social conversation than the Super Bowl commercials themselves.

Five hundred eighty thousand brand mentions in two weeks. Out-earned by a bird.

The lesson most people took from that is that earned and shared attention beats bought attention. (Here is that “unpaid media” rearing its ugly head again! If only we had a solution for that.)

This, of course, is true, but the gatekeeper era adds that attention isn’t admission, either.

Because when we ran Duolingo through the PESO Model Diagnostic, we found that the brand’s most famous asset—the best shared media engine on the planet—contributes almost nothing to how AI systems describe it.

The models don’t cite TikTok comments. (The persona doesn’t watch TikTok.)

They cite the machine-readable record, which includes coverage, reviews, structured owned content, and research corroborated by credible sources.

Duolingo’s owned authority layer is thin relative to the brand’s fame, which means the thing it does best is nearly invisible at the gate.

The brands trying to buy their way in are locked out. And the brands that only chase virality are standing outside a different door. The ones who get in are those doing the least glamorous work in this industry—publishing real expertise, earning third-party validation, keeping the story consistent everywhere it appears.

The wrong conversation is “how do we get seen?” The right question is, “What would make a machine confident enough to vouch for us?”

What to Do Now

If you’ve been running an integrated PESO Model program—or building one through the PESO Model Certification®—I have good news: you already own the skill set for this era.

You’ve been building corroboration engines your whole career, whether your business card says marketing or communications.

And if you took my advice last year and built an AI persona, even better—you already know this “audience” wants consistency, corroboration, and content it can actually read. The only thing that changed is the persona’s job description. It used to be a reader. Now it’s a referee.

Here’s where I’d focus.

First, audit your claims against your corroboration. Last week, I suggested you audit the three most important things your brand says about itself. If you haven’t already done that, I’m repeating it because it’s that important.

When you do the audit, ask for each one, where does earned media confirm this? Where is the community saying it without us in the room?

A claim that lives only on your own website isn’t evidence—it’s an assertion waiting for validation.

(The PESO Model Diagnostic scores exactly this for you—it’s free, and you can take it as often as you’d like.)

Second, re-brief your media relations program as an admission strategy. Third-party validation is no longer a warm-feelings awareness play; it’s the single most heavily weighted signal at the gate.

When a journalist covers you, a trade publication quotes your executive, or an analyst cites your data, you’re not reaching their audience anymore—you’re building the evidence file the gatekeeper reads.

Tell your team. Better yet, tell your CEO.

Third, reframe your paid strategy around amplification rather than access. Put paid dollars behind the content that already has corroboration—the covered launch, the cited research, the case study a customer will echo. Paid pointed at corroborated claims compounds your credibility. Paid pointed at uncorroborated claims is a well-lit, empty store.

And finally, take this to leadership as an opportunity. Somewhere in your organization, someone is about to propose solving “the AI problem” with a bigger ad budget.

You now get to explain—calmly, with Gartner citations and Gini snarkiness—why that’s the one lever that no longer reaches the decision-maker.

That’s not a defensive conversation. That’s the moment you become the most strategically important person in the building.

The gatekeepers are back. But this time, nobody can buy their way past you, either.

One more thing before you go build your evidence file, because I see it almost every time a brand takes this on alone… do not, do not, do not treat corroboration like a campaign.

Six weeks of earned media outreach, a flurry of executive thought leadership, a measurement dashboard nobody opens by November—and then the budget cycle reasserts itself, and everyone quietly goes back to buying reach.

But corroboration doesn’t work on a campaign clock. The gatekeeper reads your record continuously, so the record has to be kept up to date continuously, which makes this an operations problem, not a creative one, and that’s the part teams most underestimate.

(We watched this exact pattern in a recent engagement—the strategy was right on day one; the operating cadence was what needed the work.)

So if you’ve already mapped your three biggest claims against your corroboration and found the uncomfortable silence where the earned media should be, that’s a 30-minute conversation worth having.

Shoot us an email—we’ll help you figure out the first move.