Let’s start with a polite fiction the ad industry keeps telling itself, over cocktails and six-figure SaaS invoices:

“Programmatic is fine.”

No, sweetheart. It’s not.

It’s a sluggish parade of spaghetti-tagged inventory, over-engineered dashboards, and a value chain so bloated it makes the Olive Garden lunch menu look lean. You click a button and pray the right ad lands on the right eyeballs.

Spoiler: it doesn’t.

It lands on a mobile game in Latvia at 3am, flagged as “brand safe” by someone who hasn't looked at a screen since 2016.

And while most of the industry is still clapping for 50% viewability and pretending CTRs aren’t a cry for help, Jules Minvielle is over here building an actual solution.

He’s not just a founder. He’s the Co-Founder and CEO of Olyzon, a company that believes legacy DSPs aren’t just outdated—they’re obstacles. This isn’t Jules’ first rodeo either. He co-founded Mozoo, scaled Ogury into one of Europe’s most buzzed-about adtech firms, and now? He’s swinging an axe at everything those previous companies helped prop up. If that’s not personal growth, I don’t know what is.

His new play? Something called Agenic Media—and before you roll your eyes and mutter “great, another AI buzzword,” hang tight. This one actually has teeth.

It’s not just a name. It’s a total rebuild of how media gets bought. Minvielle’s version of AI doesn’t just plug into targeting. It plugs into everything: briefs, RFPs, content analysis, campaign execution, and contextual intelligence. As he puts it, “The agents are involved in everything—from the moment we receive an email from a client to delivering contextual insights.”

Let’s be real. Most adtech companies say “AI” and mean “a rule-based logic model with a buzzword attached.” But Olyzon’s tech is designed to scale real autonomy—to parse campaign briefs, normalize inputs, translate them into contextual triggers, and then cherry-pick the best content at the program level.

You know what doesn’t do that?

Your DSP.

You know what does?

Agenic agents that operate like caffeine-fueled analysts with zero bias and no PTO.

This isn’t just more automation. It’s a system that wakes up every morning, reads the entire transcript of a TV show, identifies key scenes, understands the mood, tone, setting, and themes, and then matches that content to your brand goals. Not genre. Not network. Not “family-friendly” BS. Actual semantic matching in real-time.

Jules said it himself: “We cherry-pick the right programs. That’s something a human cannot do… at scale.”

That’s not AI as lipstick. That’s AI as a scalpel, dissecting every inch of CTV’s chaotic jungle.

Jules is actually generous—he says DSPs weren’t “designed broken.” He thinks they were created with good intentions: to bring liquidity to a fragmented market. Which, sure. In the same way AOL chat rooms were once cutting-edge.

But let’s not kid ourselves. He adds, “Today it’s broken because some players have interest in letting it operate this way.” Read that again. Some players have interest. That’s the polite French way of saying, “Everyone is making too much money from inefficiency to care.”

He’s not wrong.
The ad supply chain has become a bloated cruise ship, where no one’s quite sure who’s steering, but everyone’s collecting a fee.

And while AI could streamline the entire thing, most platforms are treating it like a sticker. Jules has no patience for that. “We’re not a platform,” he says. “Platforms require a login. Our agents plug into anything.” The man is coming for your dashboards, your workflows, and probably your job title.

Let’s not pretend this is just a tech problem. Advertisers are complicit in keeping the machine running. Jules explains it simply: “If everything works, it’s okay to say, ‘I don’t want to touch anything.’” That’s not strategy. That’s cowardice in PowerPoint form.

And publishers? Still terrified of giving advertisers too much control. “Everyone is bundling inventory,” he says. “But there’s more value in unbundling and showing the true context of each show.” The fear is that buyers will cherry-pick only the shiny stuff like Emily in Paris and ignore the B-roll. Jules disagrees—“Showing the value of underappreciated programs brings in more demand, not less.”

He’s essentially saying: stop hiding your best assets behind lazy packaging.
You’re not Costco. You’re a content platform. Act like it.

This isn’t another interview fluff piece. It’s a six-part takedown of the bloated, broken, and barely-functioning structure we call programmatic. We’re going deep—into context vs cookies, signal vs scale, agentic intelligence vs lipstick AI, and how Jules would fix Netflix and cook a multi-course French meal while doing it.

And let’s be clear: Jules says “clarity wins.”
I say: clarity is brutal, and most of this industry isn’t ready to be that honest.

Over the next five parts, you’re going to hear the things nobody wants to say out loud at Cannes, but everyone whispers about over overpriced rosé. You’ll learn what real AI in media looks like. You’ll understand why most DSPs are just glorified UX prisons. And you’ll watch as the polite, espresso-fueled Frenchman quietly declares war on adtech’s sacred cows—and then serves them with a béarnaise reduction.

Stay bold. Stay curious. And know more than you did yesterday.
Or at least before your next QBR.

For Jules Minvielle, programmatic advertising hasn’t failed due to lack of intelligence—it’s failed because the system was built around the wrong incentives. That’s why he launched Olyzon, a company he describes as “AI-native from the ground up,” rather than another platform adding a chatbot to aging infrastructure.

He explains that the company was born just after the largest language models hit the market, and that timing was deliberate. The technology wasn’t simply added to what existed—it defined how the company was built. Instead of using AI to patch inefficiencies, Olyzon’s agents are foundational, operating across targeting, planning, and execution with little need for human mediation.

Minvielle describes this shift as both energizing and liberating. Though he's 40, he says he feels 25 again—not because he's romanticizing startup life, but because the technology has fundamentally changed the pace and scope of what’s possible. And unlike previous ventures—Mozoo and Ogury, both of which he co-founded—this time, he’s not building on legacy infrastructure or retrofitting teams to new tools. The company is structured for autonomy at scale.

The core innovation lies in what Minvielle calls “agents,” purpose-built AI units that interpret vague creative briefs, normalize objectives, and generate contextual media plans that are rebuilt continuously as programming changes. When asked what makes this different from standard DSP workflows, he explains that it's about both scale and specificity.

For example, a recent campaign arrived with nothing more than a short paragraph—what he calls a “five-line brief.” Traditionally, such a request would trigger a flurry of manual labor, followed by a static inclusion list based on historic assumptions. In this case, the agents extracted the implicit audience and intent from the language, mapped it against program-level data—not just genre or publisher-supplied tags—and generated a dynamic targeting set that updated daily.

The difference, according to Minvielle, is that the campaign would likely have underperformed or been delayed entirely under a traditional DSP setup. With agents, it went live on time, with targeting logic refined far beyond what a typical inclusion list could offer. He emphasizes that “a human team could theoretically do this, but not daily, not at scale, and certainly not without friction.”

This system is especially relevant in CTV, where identifiers are unstable, audiences are shared, and user-level precision is murky at best. Minvielle notes that in a household of four or five people, a single IP address or cookie can’t distinguish who’s actually watching. Context, however, can.

By analyzing transcripts and metadata, Olyzon’s agents determine whether a given program is likely being viewed by a child, a parent, or a specific interest group—information that can outperform demographic targeting. According to Minvielle, “Context beats cookies. Every time.” The company’s internal benchmarks suggest significant performance lifts in both brand recall and engagement when contextual alignment is prioritized over identity proxies.

Olyzon’s thesis is straightforward: if you know what someone is watching, you can infer why. And when that inference is powered by agents working continuously, rather than humans checking dashboards, the result isn’t just better targeting—it’s an entirely different system architecture.

The next installment will explore why that architecture may soon render most DSPs functionally obsolete—and what happens when platforms built for control are replaced by systems built for delegation.

(Or: How to Stop Reading Industry Theater and Start Seeing the Blueprint)
Because while you're sipping $8 lattes and scrolling LinkedIn jargon, the real stories are happening in DMs, leaked decks, and metadata footprints.

Let’s get one thing straight:
We’re not here to summarize press releases.
We’re not here to repost Series A announcements and call it “insight.”
And we sure as hell aren’t here to make everyone feel comfortable.

ADOTAT publishes 2,000 to 3,000 words a day—for free. That’s more than most VC-backed “media brands” cough up in a week. But the real good stuff—the 5,000 to 10,000-word deep dives that pull the skin off the adtech façade—that’s behind ADOTAT+.

Because real reporting doesn’t scale on vibes.

It scales on subscriptions.

We’re not building yachts.
We’re building infrastructure for the only independent journalism operation that’s actually reading the whitepapers, the patents, and the fine print in your vendor contracts.

Your subscription lets us:

🔬 Expand to 10,000 words per day when it matters — We don’t just run with the first pitch. We talk to the people who built it, broke it, and got fired over it.

📡 Investigate the stuff nobody else touches — Industry blacklists. Measurement fraud. The AI vendors building “attention scores” with zero peer review.

🧠 Keep adtech honest — Not because we hate it. Because we actually care if it works.

The industry is bloated with newsletters that say “curation is hot,” “retail media is rising,” “attention is the new currency.”

Cool. We said that two years ago, and we told you why, who, and who’s bluffing.

ADOTAT+ subscribers don’t just know what’s coming.
They already shipped the memo to their board.

You can keep reading LinkedIn posts by people who haven’t launched a product since the iPhone 4…
Or you can subscribe to ADOTAT+, and finally get the full story.

Because the next phase of this industry?
It’s not going to be explained in 280 characters or PR quotes.

It’ll be decoded—by us. For you. Before the conference stage catches up.

👉 [Join ADOTAT+ Now] — Before your competitors read it and start quoting us in your next strategy meeting.

🧠 Stay Bold. Stay Curious. Know More Than You Did Yesterday.
(And stop getting blindsided by your own media plan.)