So, yeah. I turned 50 today.

Fifty. Five-oh. Half a century on this floating rock, most of it spent clicking “accept all cookies” just to read news I didn’t ask for.

And somehow, through a delightful genetic fluke known as Ehlers-Danlos, my skin hasn’t gotten the memo. I still get people squinting at my ID like I Photoshopped it. If I shaved my beard, I’d be offered a summer internship.

But while my face is serving fresh-off-the-runway Sephora sample, the rest of me? It's slowly being replaced by mystery parts from a Radio Shack clearance bin. My joints creak like ad servers running on Flash. I’m the Benjamin Button of B2B—but instead of aging backward, I’m just dying politely.

At 50, I figured I’d be zipping around in a self-driving Uber drone, sipping mushroom coffee, and finally freed from the tyranny of line items.

But no. We’re still here.

Still pretending that clicking around a DSP counts as “strategy.” Still slicing impressions into fragments and calling it optimization.

Enter Agentic Media—the buzzword with bite.

Let me put it this way:

If you’re still talking about AI as something you “add to your stack,” you’ve missed the plot.

The future isn’t platforms with AI.

The future is agents instead of platforms.

And this week, two guys who helped build the industry are now gleefully pouring gasoline on it.

Jules Minvielle, co-founder and CEO of Olyzon, isn’t some deck-churning hype guy. He’s built real businesses—Mozoo, Ogury, and now a company that doesn’t believe in dashboards. As Jules puts it, the DSP wasn’t designed to be broken. It just stayed broken because everyone makes money pretending it works. Olyzon is his answer: not an optimization layer, but a full AI-native replacement that doesn't need your login credentials or your bloated inclusion list.

It reads your brief.

It reads the script of a show.

And then it decides whether your ad belongs next to it.

Not based on genre. Not based on “family-friendly” tags from 2007.
Based on mood. On theme. On whether the protagonist is experiencing character growth or just yelling about cryptocurrency.

Meanwhile, you’ve got Brian O’Kelley over at Scope3, who helped invent the modern ad exchange and is now trying to dismantle it like someone pulling their name off a bad group project. He didn’t just bring agentic intelligence to brand safety—he dragged it into every corner of digital buying. He said flat-out: we don’t need brand safety tools that operate like blunt medieval flails. The problem with old-school categorization is that it blocks “death of a star” and “death of a character” in the same breath.

That’s not safety.

That’s stupidity with a badge.

Agentic AI isn’t about controlling media. It’s about delegating it.
It doesn’t need you to log in. It doesn’t need you to watch dashboards like a helicopter parent. It needs clear goals—and then it does the work for you.

Jules built agents that read the campaign brief, understand context, and then cherry-pick the exact scenes in streaming content where your ad will make the most sense. He literally said that a human couldn’t do what his system does—not because we’re lazy (though we are), but because it would take a thousand interns and an IV drip of Adderall to process that much data every single day.

Brian, on the other hand, is over here asking questions most vendors won’t touch. Like, why are we still building brand safety models based on keyword paranoia and 2006-era taxonomy trees? Why is every point solution a copy-paste job with “AI” added to the title like it’s parsley?

He doesn’t want to fix the dashboard. He wants to eliminate the need for one entirely.

Let’s strip it down. It’s not just an “AI-powered” tool bolted onto an old UI.
It’s an autonomous system that reads, interprets, adapts, and acts—without waiting for a media planner to wake up, sip burnt Keurig coffee, and click “optimize.”

We’re talking full autonomy:

Jules calls it cherry-picking shows with surgical precision.
Brian calls it burning ad tech to the ground and rebuilding it with logic and sustainability.
I call it… about freaking time.

If you're a marketer still asking your team to pull weekly Excel exports from a DSP UI that looks like Windows XP, you’re already behind. Your competition is delegating their media strategy to systems that don’t need caffeine or PTO.

And here’s the brutal part: they’re doing it better than you could, faster than you can, and with less bias, friction, or legacy fluff.

Agentic AI isn’t the next trend. It’s the new foundation.
The systems being built by Jules and Brian don’t optimize within the old rules. They change the rules entirely. They don’t require you to approve bids—they don’t even wait for you to notice.

And good riddance.

The question isn’t if agentic AI will replace your legacy ad tech stack.
It’s how long you’re going to pretend your spaghetti-tagged DSP is still working while your competitors let agents make smarter decisions before breakfast.

So light a candle. Say kaddish for the dashboard.
And let the agents take it from here.

🧠 Stay bold. Stay curious. Know more than you did yesterday.
(Or keep logging in and pretending you’re in control.)

Linthicum told us companies are “agentic-washing” their products—tacking on the label without delivering the functionality. It’s all sizzle, no infrastructure.

In his view, only a sliver of enterprise problems (around 5%) are actually suited for agentic AI. The rest? Overkill. Expensive, complicated, and unlikely to deliver ROI.

Most implementations rely on GPU-heavy stacks, costing 5–10x more than traditional systems. Unless you're made of NVIDIA stock, good luck scaling.

The industry is confusing “assistive” AI with true autonomy. Most systems still require a human adult in the room—despite what the sales decks suggest.

He compared it to the early cloud days—massive overspend followed by a wave of repatriation and regret. Agentic AI could be next.

Linthicum teaches one of the top LinkedIn Learning courses on agentic AI. But if you’re applying the tech to the wrong problem, he says you’re not innovating—you’re just setting money on fire.

David Linthicum wants you to pause before buying the hype. Agentic AI has real potential—but without use cases, infrastructure, and pragmatism, it’s just another moonshot torching your budget.

Once upon a time—not so long ago, but far enough to now feel naïve—the entire promise of digital advertising could be summarized with one word: control. Marketers were told they could log into sleek dashboards, toggle hundreds of levers, set their targeting like a maestro tuning an orchestra, and reach exactly who they wanted, exactly when they wanted, in exactly the right frame of mind.

Of course, that was mostly fiction.

What the industry got instead was complexity masquerading as control—a labyrinth of dropdowns, filters, and campaign settings that turned every media buyer into an unpaid systems administrator. Behind the curtain, it was often just the same recycled inventory and cookie-based guesswork shuffled through prettier interfaces.

And this is precisely where the shift to agentic media begins: not with the question of how to regain control, but whether “control” ever delivered what it promised in the first place.

Demand-side platforms weren’t engineered for streaming video, podcast midrolls, or TikToks that go viral before breakfast and disappear before dinner. They were designed in an era of static websites and desktop display banners. Their taxonomy-driven structures—grouping content into rigid categories like “sports” or “fashion”—assumed a stable, categorizable internet.

But we don’t consume media like that anymore. Not even close.

Enter Jules Minvielle, whose company Olyzon skips taxonomies altogether and goes straight to the source: transcripts. His platform reads entire pieces of content—podcasts, videos, streaming episodes—understands their actual subject matter, sentiment, and context, and then determines in real time whether your ad belongs there. The future, in his view, isn’t about finding the right category. It’s about understanding the conversation as it’s happening.

Brian O’Kelley, known as the architect of modern programmatic, now spends his time setting fire to the house he helped build. His assessment of the current ecosystem is blunt: “Most of adtech is vapor wrapped in dashboards.”

To put it differently: the tools we thought were sophisticated were often just cosmetic. They looked like trading terminals but operated more like slot machines. Behind the toggles and targeting were systems still reliant on outdated heuristics, mismatched taxonomies, and unreliable identity graphs.

O’Kelley’s solution? With Scope3, he’s building a system that doesn’t just tweak the inputs—it replaces the entire concept of bidding. Scope3’s platform uses autonomous AI agents that consume environmental, contextual, and content data to make real-time decisions—no human intervention required.

This is not optimization. It’s displacement.

The real revelation here isn’t technological—it’s psychological. For years, the illusion of control was the primary value proposition in media buying. The ability to click, customize, adjust. But now, marketers are discovering that the best outcomes don’t come from fiddling with dials—they come from trusting systems built to learn faster than humans can tweak.

Consider Butler/Till, a media agency that tested this theory by ceding manual oversight to autonomous AI agents. The result? A 7% lift in brand suitability—not by micromanaging placements, but by letting go. Their success underscores a key point: autonomy is no longer a compromise; it’s a performance strategy.

Let’s get granular. Why are transcripts—actual word-for-word content records—such a game changer?

The dashboard promised you visibility, but visibility isn’t strategy. It isn’t even insight unless it’s actionable—and most dashboards weren’t. They were elaborate rituals. Daily logins to pull reports, tweak CPM floors, and pretend that adjusting bids by $0.03 was a bold strategic move.

Minvielle puts it plainly: “Platforms require a login. Agents just plug in.”
That’s not just clever—it’s a blueprint.

The next wave of adtech won’t need you to babysit it. It will listen, decide, and act—all without demanding that you spend half your day clicking around like a confused tourist in Excel.

This transition isn’t just useful for marketers and media buyers. It’s a case study in how entire toolsets—and the habits around them—can become obsolete.

In the classroom, the rise of agentic media provides a masterclass in:

You don’t need a consultant to explain this to you. You need to test it, observe it, and iterate.

Here’s a quick-start:

Bottom Line:
The era of logins, taxonomies, and dashboards is winding down. The new frontier is about delegation—not in theory, but in actual deployed software that reads, thinks, and acts.

You don’t need to control everything.
You need to define your objectives, then step back.

The machines are ready.
The only question is: are you?

👉 Next: Inside ADOTAT+ — case studies, transcript-based vendor comparisons, and how this shift is rewriting staffing models across holding companies and independent agencies alike.

Think this was spicy? That was just the free sample. In ADOTAT+, we don’t just name the problems—we trace the revenue trails, decode the jargon, and show receipts. You’ll get the 10,000-word deep dives the trade press is too polite (or too sponsored) to publish. We break down who’s faking AI, which vendors are on life support, and how agentic systems are rewriting adtech’s power map.

👉 Subscribe and get the unfiltered briefs, real market shifts, and the backchannel intel the insiders already read.

No middlemen. No dashboards. Just truth.

[Subscribe Now — Before the Middle Eats Itself]