AI agents entering programmatic ad buying
Ad-tech is moving from concept to practice: AI agents that can make ad-buying decisions are surfacing in programmatic markets, with the IAB Tech Lab creating an Agent Registry and agencies testing agentic SSP tools. Early pilots, including work by Wpromote and a closed beta from Kargo, show advertisers are experimenting with automated bidding and optimization as part of campaign stacks. That trend shifts true analyst work toward data hygiene, experiment design, and explaining model decisions rather than manual bid adjustments. ( )
A software agent is starting to do one of advertising’s most mechanical jobs: deciding which digital ad impressions to buy, how much to bid, and when to change course mid-campaign. In April 2026, that idea moved from conference-talk material into live tests, as standards bodies and ad-tech companies began putting agentic buying tools into real programmatic workflows. (adage.com) Programmatic advertising is the system that buys ad space through automated auctions that run in fractions of a second. When a person opens a webpage or app, software evaluates that impression, checks whether it fits an advertiser’s target, and places a bid before the page finishes loading. (adage.com) For years, that process was already automated, but the automation was narrow. Human traders still set many of the rules, adjusted pacing, watched performance dashboards, and made repeated bid changes across platforms when results drifted. (adage.com) An artificial intelligence agent changes that division of labor by acting more like a junior operator than a fixed script. Instead of following one static rule, it can take a goal such as lower cost per acquisition or stronger reach in a certain audience segment, evaluate incoming data, and make a sequence of decisions to pursue that goal. (iabtechlab.com) That creates a new problem: trust. If one company says its agent can buy media, another company needs a standard way to know what that agent is allowed to do, what data it can access, and whether it can safely connect to other systems in the ad market. (iabtechlab.com) The Interactive Advertising Bureau Tech Lab is trying to solve that plumbing problem before the market gets messy. Its new Agent Registry is a directory where companies can register advertising agents so partners can discover them and review their capabilities inside a common framework. (iabtechlab.com) The registry sits inside a broader standards effort called Agentic Advertising Management Protocols, which the Interactive Advertising Bureau Tech Lab describes as the umbrella for its agentic work. The group has also been developing an Agentic Real-Time Bidding Framework so these tools can plug into existing ad-buying pipes instead of forcing the industry to rebuild everything from scratch. (iabtechlab.com; adexchanger.com) The practical push is coming from agencies and platforms that want fewer screens and fewer manual handoffs. Ad Age reported on April 7, 2026, that ad-tech providers are pitching agentic systems that can make media-buying decisions, with the registry serving as a vetted menu of tools for the market. (adage.com) One of the clearest early examples came from Kargo, a sell-side platform and advertising company that launched Project Kera, a chat-based planning and buying environment. AdExchanger reported on April 7, 2026, that the product was in closed beta with Hershey, travel media network Navigator, and independent agency Wpromote testing full campaigns. (adexchanger.com) Kargo’s pitch is not just a prettier interface. Chief executive Harry Kargman told AdExchanger the goal is to aggregate biddable and non-biddable inventory into a single platform, which would let a buyer work through one system instead of splitting decisions across separate channels and tools. (adexchanger.com) Wpromote’s role in those tests matters because agencies sit close to the day-to-day mess of campaign execution. If an agency is willing to let an agentic tool handle more of the bidding and optimization loop, that suggests the category is moving beyond demos and into tasks that used to belong to human traders. (adexchanger.com; adage.com) That does not mean media teams disappear. It means the job shifts upward, away from constant knob-turning and toward cleaner inputs, tighter testing, and better explanations of why a model made a decision. Ad Age described the change as a move toward data hygiene, experiment design, and interpreting model behavior rather than manually adjusting bids all day. (adage.com) The near-term reality is likely to be partial autonomy, not fully hands-off buying. The Interactive Advertising Bureau Tech Lab has framed agentic execution as something the market can layer onto existing standards, which implies a transition period where human buyers, platform rules, and software agents all share control. (iabtechlab.com; iabtechlab.com) What changed this spring is that the conversation stopped being purely theoretical. Between the Interactive Advertising Bureau Tech Lab’s registry and standards work, and live pilots from companies such as Kargo with Wpromote and Hershey, agentic ad buying is no longer a futuristic slogan; it is becoming another layer in the programmatic stack. (iabtechlab.com; adexchanger.com; adage.com)
