AI Shopping Agents Disrupt Traditional Advertising Models
What happens to advertising when the buyer is an agent is that almost everything in digital advertising assumes a person on the other side of the screen, but when that person is replaced by software, the entire chain is disrupted.
A person sees an ad, feels something, clicks and buys, and the ad takes the credit. However, when an agent does the comparing, the click path folds into the agent, and the credit model breaks. There are three possible futures for AI in shopping: AI as an ad surface, AI as a shortlister, and AI as the buyer.
Three Futures for AI in Shopping
In the first scenario, AI is an ad surface, where a person asks a chatbot for recommendations and sees labeled ads next to the answer. In the second, AI is a shortlister, where software compares options and recommends a few, and the person picks and pays. The third scenario, where AI is the buyer, is where the agent compares, chooses, and checks out for the person, and this is where the money questions start.
Ads already run in the first two scenarios, but the third has no proven ad model yet. The impression, which is the foundation of advertising, breaks when software does the reading. Amazon made this point in its lawsuit against Perplexity’s shopping agent, arguing that it has to detect and filter out “non-human impressions” before it bills advertisers.
Read Also: PASIKA warns VAT cut useless without price monitoring
The creative also changes, as an agent does not watch videos but reads product data, such as price, delivery date, and reviews. The product feed becomes the most important creative a brand has, and marketers will need new questions, such as are they found at all, where do they sit in the answer, and what converts after an AI referral?
Advertising to an Agent
There is an awkward problem underneath all of this: if an agent works for the buyer, paying to change what it recommends looks a lot like prompt injection with a media budget. There are three models for advertising to an agent: OpenAI keeps ads beside the answer, Amazon and Google place sponsored products inside their assistants, and the third model has no ads at all, where the buyer pays for the agent, and brands compete on data, price, and proof.
The author’s view is that if the agent works for the buyer, the ad has to work for the buyer too. An agent will pick the brands it can read, trust, and verify, and being readable means product, stock, and delivery data that an agent can call through an API or a clean feed.
Being trusted is where brand comes back in, as an agent still has to decide whom to trust, and it does that by checking reviews, policies, track record, and whether the facts are consistent. The money will move through agencies, and whoever closes the gap in reporting and proving the effectiveness of ads will get the first serious budgets.
Read Also: Winter sports options abound in Cyprus
Early moves point to this, with Google running Performance Max and AI Max campaigns inside AI Mode, and select US advertisers able to extend their Amazon DSP campaigns into ChatGPT. Cypriot businesses can take steps now, such as checking their product feed and tracking visits from AI assistants, to prepare for the future of AI shopping.
By taking these steps, businesses can ensure that they are found and trusted by AI agents. The businesses that prepare their data now will be the ones that agents can find and trust, and they will be the ones that thrive in the future of AI shopping, with Visa’s European transactions and protocols such as OpenAI and Stripe’s Agentic Commerce Protocol and Google’s Agent Payments Protocol (AP2) already live in Europe.
Preparing for the Future of AI Shopping
Another important step that businesses can take is to track visits from AI assistants as a separate channel, rather than letting them disappear into organic traffic. This will allow businesses to better understand the impact of AI agents on their sales and make more informed decisions about their advertising strategies.
