For years, marketers have been trying to influence what ChatGPT, Gemini, Claude and other AI platforms say about their brands, adapting to the rapidly changing rules of answer engine optimisation (AEO) and generative engine optimisation (GEO).
Now, some are taking the idea a step further: advertising directly to the AI agents themselves.
Ad-tech company Mobian has begun testing what it calls “Agent Ads” with publisher TIME, with Ally Bank and Project Management Institute among the first advertisers. Instead of conventional display creative, the ads consist of structured, sponsored text containing brand information designed to be consumed by AI systems as they process publisher content.
TIME has been creating stripped-down, machine-readable versions of its webpages in Markdown, removing much of the design and imagery found on pages intended for human readers. Mobian inserts the sponsored material into these agent-facing versions, generally in FAQ-style formats containing brand facts and messaging. The ads are explicitly labelled as sponsored content.
The new audience
The experiment reflects a fundamental change in the advertising audience. The immediate recipient of the message is no longer necessarily a consumer; it can be an AI system that is researching, comparing and potentially recommending products on that consumer’s behalf.
Mobian says its research involving more than 750 brands found that roughly 17% of the sources cited by AI engines conflicted with the brands’ own facts or positioning. The company argues that Agent Ads can give brands a controlled, cited source of information at the point where an AI system is gathering information.
For Ally, the experiment is particularly significant because AI referrals are already showing signs of commercial value. Ally chief marketing and public relations officer Andrea Brimmer told Mobian that consumers arriving at Ally’s website from AI platforms were 3.5 times more likely to open an account than those coming from traditional search, while AI-driven traffic had risen ninefold year over year.
The implications extend beyond individual ad formats. Expedia Group has established a dedicated business-to-agent (B2A) marketing capability aimed at influencing how AI agents discover and evaluate the travel company.
Expedia has described AI agents as a new audience alongside consumers and businesses. Its own careers material says the company is building systems to make its supply, pricing and brand “legible, accessible, and preferred” to AI agents across the travel funnel.
The move is driven by a recognition that AI agents do not necessarily behave like human customers. A traveler might choose Expedia because of familiarity or loyalty. An agent, by contrast, can evaluate options according to price, availability, preferences, benefits and other structured signals.
Expedia says B2A marketing therefore requires richer, more structured and relevant information that AI systems can understand and potentially surface to travelers.
The stakes are likely to rise as AI becomes part of the transaction itself. Expedia’s April 2026 research, based on more than 5,700 adults in the US, UK and India, found that almost 70% of travelers preferred booking with trusted travel brands over AI chatbots and agents, even as AI increasingly influences trip planning.
For marketers, this creates a new battleground. GEO asks how a brand can appear in an AI-generated answer. Agent advertising asks a more provocative question: what if the advertiser could address the AI making the answer before the consumer ever sees it?
The experiment is still in its early stages, and there is no guarantee that sponsored information encountered by an AI crawler will influence a chatbot’s eventual recommendation. But the direction is clear: as agents become intermediaries between brands and consumers, marketers are beginning to treat the machines themselves as an audience.
