As we look ahead to 2027, agentic AI is already transforming how the advertising industry operates. Using natural-language UIs, agents can help marketers plan campaigns, make sense of data, and surface optimization opportunities. This reduces manual work and, ideally, gives teams more time to focus on strategy. 

As AI agents take on more campaign management tasks, a question emerges: If an agent can orchestrate campaigns autonomously, do advertisers still even need a demand-side platform (DSP)? 

The answer is yes. 

AI agents and DSPs serve different purposes

AI agents and DSPs serve different but complementary roles. An agent can help determine what a campaign should do and coordinate actions around that, but a DSP is where that campaign meets the realities of programmatic buying. 

The DSP’s bidder evaluates auction opportunities and makes decisions about bid value, budget pacing, and frequency. These continuous, auction-level functions shape how media is bought, and how a campaign performs. They aren’t administrative tasks that can easily be moved into a chat box. 

This is an important distinction for a few reasons, not least of all because advertising is more than an exercise in issuing instructions. It’s about making decisions in a dynamic marketplace, impression by impression, while juggling the constraints and goals of a campaign. And the DSP provides the infrastructure to act on those decisions in an auction.

Why measurement still depends on the DSP

In addition to managing media buying, the DSP also supports measurement and experimentation that rely on control over bidding. For example, incrementality testing allows the bidder to withhold selected impressions to establish a control group and assess whether advertising contributed to a given outcome. This kind of test-and-learn capability requires the buying platform to participate directly in the auction. An agent, ad server, or supply-side platform (SSP) alone can’t provide the same control. 

Data platforms are equally important here, as they give advertisers a way to manage and activate their own data (and tap into a platform’s proprietary data). Together, a DSP and data platform provide the execution and intelligence that make agentic tools more useful. That’s exactly why we should look at these tools as a way to extend the value of these platforms rather than replace them.

Human judgement continues to matter

Human judgement remains essential. Automation can organize options and accelerate routine tasks, but marketers bring the experience needed to decide which opportunities are right for a particular advertiser, audience, or objective. The goal should not be to remove that judgment, but to give teams better data, information, and more room to use their expertise.

As our industry experiments with new agents and interfaces, the way we interact with ad tech will likely become more flexible and, hopefully, more intuitive. But the underlying need to value impressions, manage budgets, and run controlled tests doesn’t go away. In the agentic era, the DSP is still the engine of execution; AI just makes it smarter and easier to use.

—

Oscar Rondon is a digital media and programmatic technology leader with nearly two decades of experience in advertising, advanced TV and measurement strategy. At Nexxen, Oscar leads sales enablement and the TV data business, working directly with brands and agencies on media evaluation, planning, and performance measurement across the programmatic ecosystem.