By Katie McAdams, Chief Marketing & Commercial Officer, Basis
AI doesn’t have opinions about how advertising works. It inherits ours.
Within a year or two, an autonomous agent will move a real share of your budget without asking permission. But what it does with that money will depend on assumptions most teams have never examined. Our company’s survey of marketers show that nearly 60% of agency professionals now use AI every day—up from just 16% two years ago—and two-thirds of media buyers say they are actively planning for agentic AI to handle buying and execution.
The machines are coming, but what logic they will use?
Automation does not improve a flawed process. It scales it. An agent executing at machine speed will pursue whatever objective it inherits, across whatever map of the market it is handed, and it will do so relentlessly—and perhaps incorrectly, if the map is wrong. Before we hand over the keys, we have to correct the assumptions the industry spent a decade building into its tools. Four stand out in particular, and while each may seem like a small error today, they’ll get a lot more expensive the moment your agents start acting on them.
Myth #1: The DSP is the center of advertising
For more than a decade, the demand-side platform has sat at the center of the digital media universe, on the theory that all media would eventually be bought programmatically. The theory was half right: Roughly nine in 10 digital display dollars now move programmatically. But a buying tool is not a brain. A DSP executes decisions—it does not actively “make” them on your behalf—and it sees only the inventory it transacts.
As agents take over execution, the DSP becomes one set of hands among many, rather than the place your strategy lives. Organizing a media operation around its buying platform is like organizing a company around its checkout counter. The center of gravity belongs to the layer that connects planning, buying, measurement, and money across every channel.
Myth #2: AI can run advertising on its own
The appeal is obvious: Point an autonomous system at your accounts and let it optimize the whole thing. But buyers themselves remain cautious: many still find agentic tools interesting rather than urgent, and standards bodies openly expect false starts.
And they are right to be careful. An unconstrained agent is just a junior buyer with no budget limits, no context, and no supervisor, and who’s moving thousands of times faster.
To act safely, an agent needs guardrails, spending limits, and a live connection to what the business has actually committed to—the contracts, budgets, and obligations that define what “good” even means. Strong AI outcomes require strong inputs. The output is capped by the rails around it.
Myth #3: The campaign is the unit of media
Our industry has been trained to treat the campaign as the basic object of the work—plan it, launch it, optimize it, report on it, repeat. But the campaign is an output, not an operating system. Running the business campaign-by-campaign, channel-by-channel, is the same fragmented habit that falls apart the moment AI is orchestrating across the entire portfolio at once.
Agents do not manage discrete flights; they run continuously, against outcomes. The unit that actually matters is the whole media operation and the commitments that govern it—what you have promised, what you are spending, what you owe and are owed. Anchor the system there, and the campaigns take care of themselves.
Myth #4: Programmatic platforms are omnichannel
Nearly every platform now markets itself as omnichannel, implying unified reach and optimization. The market says otherwise. Data from IAB shows that ad spend is splintering—connected TV up 13.8%, commerce media up 12.1%, and social up 14.6% in 2026 alone—and no single platform contains it all.
Most platforms optimize their own inventory and their own scoreboard, and “omnichannel” inside one seller’s tool typically means “every channel we happen to sell.” Genuine cross-channel optimization happens above any single platform, which is exactly why cross-platform measurement became a top priority for 72% of buyers this year, up from 64%, according to our marketer survey. In truth, omnichannel isn’t a feature you can buy from any single programmatic platform. It’s a vantage point you have to own above all of them—and one that you need an operating system to implement.
Advertising in the AI era
The four myths share one root: Each puts a tool or a tactic where a system should be. Programmatic earned its place as a key component of modern media, but it does not work as an organizing principle—especially in the dawning AI era.
That distinction is about to get costly. The work of the next 18 months is not bolting AI onto the old map. It’s about redrawing that map so that when the agents arrive, they can optimize toward the whole business, across the whole ecosystem, inside real guardrails.
Global ad spend will cross a trillion dollars this year. The teams that win the next decade will not be the ones with the most AI. They will be the ones whose AI runs on the truest picture of how their media actually works. Take the time to audit your assumptions before you automate them. Because automation will not fix flawed logic. It will scale it.
