The programmatic supply chain has always been a polite fiction. Brands pay $1, somewhere between $0.40 and $0.60 actually reaches a real ad impression, and everyone in the middle calls it “efficiency.” AI agents dismantling that chain sounds like a win, and it could be, but only for brands that have done the unglamorous infrastructure work first.
That’s the part nobody is talking about loudly enough.
The Programmatic Middleman Problem Is Real And Expensive
The ANA’s landmark supply chain transparency study put a number on what performance media practitioners already knew: up to 50% of programmatic dollars vanish before they reach a working media impression. DSPs, SSPs, data brokers, verification layers, brand safety vendors: each takes a cut, each adds latency, and most add noise rather than signal.
For CPG and DTC brands spending $1M–$20M annually in media, that isn’t a rounding error. That’s your growth budget quietly walking out the door.
The holdcos built their empires on top of this complexity. They didn’t design the waste, but they got very comfortable with it.
What AI Agents Actually Change (and What They Don’t)
AI agents capable of executing media buys directly, without routing through layers of ad tech middlemen, represent a genuine structural shift. In theory, a brand’s AI could query publisher inventory directly, make a bid decision, and serve an impression without a single intermediary touching the transaction.
Cleaner. Faster. More accountable.
But here’s the part that gets glossed over: AI agents don’t manufacture data. They execute against it. If your first-party data foundation is weak, you haven’t cut out the middleman. You’ve just replaced a human making bad decisions with a machine making them faster.
What Has to Be True Before You Can Benefit
This is where CPG and DTC brands need to do honest self-assessment. Before AI-driven direct buying delivers on its promise, four things have to be in place:
1. A real first-party data asset. Not just an email list. We mean a structured, consented, signal-rich understanding of your buyers — purchase behavior, frequency, channel preference, lifecycle stage.
2. Clean identity infrastructure. Your customer data has to be resolvable across environments. If you can’t match a buyer across touch points, your AI agent is flying blind.
3. An outcome-based measurement framework. If you’re still optimizing to CTR or last-click ROAS, an AI agent will just optimize those broken metrics faster. Know what you’re actually trying to move: trial rate, repeat purchase, category penetration.
4. Media strategy that isn’t channel-dependent. AI agents are most powerful when the brand has a clear point of view on audience, message, and moment — and the technology executes that vision. Strategy still has to come from humans.
Without these four foundations, the efficiency gains from cutting programmatic middlemen largely flow to the technology vendor, not the brand.
What This Means for Independent Brands Specifically
Large holdco clients will get AI-agent media buying rolled out to them as a managed service with a new margin layer attached. That’s how it works. The efficiency savings will be partially real and partially redistributed upward.
Independent CPG and DTC brands, the ones working with focused, accountable agencies, have a different opportunity. They can build the right infrastructure now, move faster than their category incumbents, and actually capture the working media improvement that AI-direct buying promises.
Our performance media practice is already building measurement and data frameworks designed for this transition. Not because it’s trendy, but because our clients can’t afford to lose another 40 cents on every dollar they spend.
The brands that will win aren’t the ones who wait for their agency to hand them an AI product. They’re the ones who treat first-party data as a strategic asset today and choose media strategy partners who think the same way.
FAQ: AI Agents, Programmatic, and What It Means for Your Brand
What are AI agents in media buying?
AI agents are software systems that can autonomously execute media buying decisions, researching inventory, evaluating bids, and serving ads, without requiring a human to approve each step or an ad tech intermediary to route the transaction.
Will AI agents actually reduce programmatic waste for CPG brands?
They can, but only if the brand has first-party data infrastructure and clear outcome-based measurement in place. AI agents execute against the data and objectives they’re given. Weak inputs produce efficient waste, not efficient results.
What are programmatic middlemen and why do they matter?
Programmatic middlemen are the DSPs, SSPs, data brokers, and verification vendors that sit between a brand’s media budget and an actual ad impression. Each layer takes a fee. Industry research suggests only $0.40–$0.60 of every programmatic dollar reaches a working impression.
Do I need to be a large brand to take advantage of this shift?
No. In fact, independent CPG and DTC brands that build the right data and measurement foundation now have a structural advantage. They can move faster and capture more of the efficiency gains than brands locked into bloated holdco relationships.
Ready to build the media infrastructure that makes AI-driven buying actually work for your brand?
Talk to Junction 37 about performance media →
Chris Pyne, Founder, Junction 37 – 30+ Years in Performance Media.