Every single day, someone on LinkedIn posts a breathless take about how AI agents are going to revolutionize media buying. Every. Single. Day.

CES 2026 was wall-to-wall agentic. IAB ALM was worse. I stood in that room and watched executive after executive grab me by the arm, eyes shining with the fervor of the recently converted, telling me they were "doubling down on AI." They were on the side of this protocol or that framework, and I should be too. I looked at most of them blankly. Not because I don't care about the technology. I do. But because I've seen this movie before, and I remember how it ends.

It ends with consultants getting paid.

That's the thing nobody wants to say out loud, so I'll say it: the reason this all gets made so maddeningly complex isn't because the underlying technology is that complicated. It's because complexity is a business model. We figured out that most of the programmatic system doesn't actually work the way anyone promised. That somewhere around fifty cents of every dollar vanishes into the supply chain. That we somehow need "training" and "certification" to use systems that were supposed to be automatic.

And now they're doing it again. Not necessarily on purpose. But the incentive structure is the same. Take something that should be straightforward, wrap it in enough acronyms and competing frameworks and governance bodies and working groups, and suddenly everyone needs help understanding it. Suddenly there are conferences to attend and consultants to hire and advisory boards to sit on.

So here's what we're going to do. We're going to break it down. Make it simple. Because it is simple.

Strip away every deck, every keynote, every LinkedIn post with the word "transformational" in it, and here's the core idea:

Instead of humans manually setting up campaigns, pulling levers in DSP interfaces, negotiating insertion orders over email, and stitching together reports from fourteen different dashboards, you let AI agents handle the grunt work.

These agents can interpret objectives in plain English. "Find me video inventory that indexes highly among in-market electric car buyers." Then they go discover inventory, negotiate terms, execute buys, and optimize in flight. No dropdown menus. No three-day turnaround on an RFP.

That's it. That's the whole pitch. And honestly? It's a good pitch. The advertising supply chain is a Rube Goldberg machine held together with duct tape and deal IDs. An AI that can reason across all of those broken systems and close the gaps? Genuinely useful.

But here's the part they leave out of the keynote: agentic AI doesn't change the fundamental economics of digital advertising. It makes the existing machinery run faster. That's it. A new layer on top of the same plumbing.

"Faster" is not inherently good. I want you to hold that thought for the rest of this series, because everyone selling you agentic is treating speed as an unqualified positive.

Imagine giving your five-year-old a Red Bull. That kid is going to move faster, all right. Incredible velocity. Zero judgment. Bouncing off walls, knocking over furniture, making decisions at a pace that no adult can supervise.

That's what happens when you add an agentic layer to a programmatic ecosystem that's already riddled with fraud, opacity, and misaligned incentives. You don't get a smarter system. You get a system that makes the same mistakes faster and at greater scale. Without proper standards, AI agents hallucinate. They compound errors. Audiences make no sense. Placements get misrepresented. Impression goals become budgets. Budgets become impression goals.

Speed without standards is just more expensive chaos.

If you've been in ad tech for more than a couple of cycles, you've seen this exact movie. A genuinely useful technology emerges. Vendors pile on. Standards bodies draw lines. Competing camps form. LinkedIn becomes unbearable.

We did this with header bidding. Identity resolution. Clean rooms. Supply path optimization. Every shiny object since the first ad server went live.

The pattern is always the same: genuine innovation gets buried under a mountain of marketing decks, acronyms, and turf wars. The tech eventually matures. Things get better incrementally, not revolutionarily. The people who got rich were the ones who sold shovels during the gold rush.

This is programmatic doing its thing. That's all. Don't be scared. Don't let anyone make you feel like you're falling behind because you can't recite the difference between AdCP and ARTF from memory. You're not falling behind. The people panicking are either selling something or haven't been around long enough to recognize the cycle.

For all the noise, the entire agentic advertising debate boils down to two approaches. Just two. Everything else is commentary, political positioning, or LinkedIn performance art.

Camp 1: AdCP (Ad Context Protocol). Backed by Scope3, PubMatic, Triton Digital, Yahoo, Optable, and Swivel. Governed by the new Agentic Advertising Organization (AAO), with Brian O'Kelley and former IAB CEO Randall Rothenberg on the interim board. Their pitch: build something agent-native from the ground up. A new shared language for how AI agents communicate intent, constraints, and reasoning. Not a replacement for RTB, they insist, but a strategic layer above it. The biggest proof point so far: Prebid just launched the Prebid Sales Agent, open-source, built on AdCP, free on GitHub.

Camp 2: ARTF (Agentic RTB Framework). IAB Tech Lab's play, led by CEO Anthony Katsur. Their pitch: the industry already has battle-tested standards refined through billions of transactions. OpenRTB. AdCOM. OpenDirect. They all share the same underlying object model. A video impression means the same thing whether you're running an auction or setting up a direct deal. Don't torch a decade of infrastructure. Extend it. Tech Lab has committed to making all its standards agentic by end of 2026.

Both have a point. Neither is going to kill you.

Here's where it gets uncomfortable, and where you could see the discomfort in real time.

I stood in the room at IAB ALM and watched it happen. Someone asks the obvious question during a session and you can literally watch the blood rush to an executive's face as they scramble for the answer.

I watched Index Exchange's Canadian leadership turn beet red trying to explain to a room full of people why you would still need them when the robots are doing all the work. The audience wasn't hostile. They were just... confused. Genuinely confused. If AI agents can discover inventory directly, negotiate terms, and execute buys, what exactly is the exchange doing? Why am I paying you a percentage of my media spend to facilitate a conversation you're no longer part of?

"Why not just buy directly then?" someone asked. It hung in the air like a fart in an elevator. No good answer came.

This is the existential question for every ad tech intermediary, and the honest answer is: they don't know yet. The diplomatic answer you'll hear at every conference involves "value-added services" and "governance" and "trust infrastructure." Some of that is real. Someone has to provide the compute, the fraud detection, the billing, the compliance layer. But a lot of the current margin stack is going to get compressed, and everyone in the supply chain knows it.

That's why the standards fight is so vicious. It's not really about whether agents should use new schemas or old ones. It's about who controls the rails. Whoever sets the standard sets the terms of participation. That's worth billions.

Now let me bring you back to earth.

The actual tests happening right now are tiny. MiQ is running AdCP tests in the five-figure range. Teqblaze is working with about a dozen publishers. Scope3 is trying to make publisher catalogs discoverable. These are pilot programs, not market transformations.

Buyer interest is basically zero. One programmatic director at a major agency said it plainly: no client-side interest because it's too early. The current state of agentic is nice material for conference presentations. That's it.

Publishers are curious but lost. The sell-side doesn't have a clear understanding of the benefits yet. Many want to explore agentic buying but don't know what to expect and don't have the technical chops to participate.

None of this means agentic is fake. The technology is real, the use cases are real, the efficiency gains will eventually be real. But the timeline everyone is selling you is fiction. We are in the "interesting conversation" phase, not the "restructure your business" phase.

Three things:

One. Agentic advertising is a new automation layer on top of programmatic. It is not a revolution. It is not going to kill your DSP tomorrow. It is going to make certain workflows faster and cheaper over the next two to three years. That's good. That's also boring, which is why nobody frames it that way.

Two. There's a standards fight between AdCP and ARTF. It matters, but not to you right now. The likely outcome: messy coexistence for 18 to 24 months, then convergence. This is how it always works. Don't pick a side. Don't panic. Watch.

Three. The real disruption isn't the AI. It's the disintermediation. If agents can connect buyers and sellers directly, the entire margin stack gets questioned. That's the conversation that makes people sweat, and it's the one worth paying attention to.

If you want to know who's actually full of it, who has the stronger technical argument, what the transparency concerns are that nobody is addressing, and what this means for your specific corner of the industry...

Parts 2 and 3 are where we stop being polite.

We name names. We trace the money. We explain what each side actually wants versus what they say they want. And we give you the practical survival guide: what to do, what to ignore, and what to watch.

ADOTAT subscribers get the inside baseball.