The Next Ad Market May Not Be Built for Humans
For most of the internet, advertising has had one basic job: get a person's attention. A banner had to be seen. A search ad had to be clicked. A sponsored product had to catch the shopper before a competitor did.
That assumption is starting to fail.
This week, several developments pointed toward something stranger. Amazon described ads that can live inside AI shopping conversations. Time is experimenting with advertising specifically designed for AI agents to read. McKinsey argued that commerce media networks may need to stop selling digital shelf space and start competing for inclusion inside AI-generated shortlists. At the same time, the systems buying and managing those ads are becoming agents themselves.
AI commerce is creating a new commercial layer between advertising and purchasing. The customer may still pay. But increasingly, a machine may decide what deserves the customer's attention first.
Amazon Is Putting Advertising Inside the AI Shopping Conversation
Amazon Ads published guidance this week describing how brands can reach consumers who shop with AI agents. The important part was not another recommendation engine. Amazon described advertising formats inside Alexa for Shopping and Alexa+, including sponsored prompts, conversational ads and Alexa+ Agentic Ads, where a shopper can discover something, ask questions and complete a purchase without leaving the conversation.
Think about the difference.
A traditional ad interrupts a shopping journey. An agentic ad can become part of the journey itself. Someone asking about concert tickets could see a relevant promotion, ask Alexa about seats and prices, book them and receive the tickets without opening another interface.
That changes what advertisers are buying. The scarce asset is no longer simply an impression on a page. It is participation in a machine-mediated decision.
For brands, that makes the quality of underlying information more important. Amazon says agents combine product details, reviews, price, inventory and brand content when forming recommendations. If those signals are weak or inconsistent, a brand can disappear before the shopper sees a shortlist.
Mental Model: Decision Proximity
The closer a business can place itself to the moment a decision is made, the more valuable that position becomes. AI conversations are creating a new decision point before the traditional product page.
The most valuable advertising position may no longer be where the customer looks, but where the agent decides.
Source: Amazon Ads
Publishers Are Starting to Sell Advertising to Machines
A far more unusual advertising experiment surfaced this week.
Time and adtech company Mobian are testing Agent Ads: sponsored information designed specifically for AI crawlers rather than human readers. Business Insider reported that Time provides AI agents with stripped-down, machine-readable versions of pages. Sponsored brand information can then appear inside those versions in an FAQ-like format. Ally Bank and the Project Management Institute are among the early advertisers.
Mobian describes its product simply: ads built from verifiable brand information and placed on pages AI agents already consume.
That creates an entirely new type of media inventory.
A magazine once monetized human readership. Now the same publisher can potentially monetize the machine readership sitting between the publication and the human.
The experiment matters because AI systems increasingly retrieve information before presenting an answer. If brands can pay to place clearly identified commercial information inside sources agents consult, advertising starts moving upstream from influencing people to influencing the information environment machines use.
There is an obvious problem: trust. Business Insider reported an active industry debate over whether this becomes legitimate advertising or an attempt to manipulate AI recommendations. Time's implementation labels the material as sponsored, but standards for machine-facing advertising remain immature.
Mental Model: New Audience, New Inventory
Whenever a new audience becomes economically valuable, someone eventually creates inventory to reach it. AI agents are beginning to qualify as an audience.
Advertising no longer requires a human impression to have commercial value.
Sources: Mobian | Business Insider
Retail Media Is Being Forced to Compete for the Shortlist
For years, retail and commerce media had a powerful advantage: they owned the digital shelf closest to the transaction.
McKinsey argued this week that AI assistants are weakening that advantage. When someone asks an assistant for the best running shoes under $150 and receives three recommendations, the user may never see the retailer's search results, category page or sponsored product shelf.
The advertising problem becomes brutally simple:
Were you one of the three?
McKinsey describes the emerging scarce resource as eligibility inside systems that decide what gets considered, compared and recommended. It argues that commerce media networks will increasingly need machine-readable product data, rules governing sponsored influence and new ad products built around decision moments rather than page placements.
Consider a cereal brand paying for top placement in an online grocery search. That placement has value only if the shopper visits that search page. If an AI assistant compares nutrition, price, availability and reviews before presenting two options, the sponsored shelf was never entered.
Advertising therefore begins moving from buying visibility toward buying legitimate influence over consideration.
That distinction will matter enormously because AI answers are compressed. A webpage can display twenty products. An assistant may confidently recommend three.
Mental Model: Shortlist Economics
As the number of options presented to the buyer shrinks, the value of inclusion rises sharply. AI does not eliminate competition; it compresses where competition happens.
Owning the digital shelf matters less when the buyer never visits the aisle.
Source: McKinsey
The Advertiser Is Becoming a Machine Too
The other side of the market is changing at the same time.
Amazon Ads also detailed a connected AI advertising workflow this week. Its Creative Agent can generate campaign assets. Ads Agent can handle activities including targeting, bidding and campaign launch. An Amazon Ads MCP server allows external AI agents to interact with Amazon's advertising API through natural-language instructions.
That creates an important second-order shift.
We are moving toward an advertising market where AI may exist on both sides of the transaction.
One agent could decide how a brand allocates media budget, generate creative and optimize a campaign. Another agent could evaluate products for a consumer. Between them sits an advertising system trying to determine which commercial message deserves influence.
Humans still define objectives, budgets and brand boundaries. But more of the continuous negotiation underneath advertising can become machine-operated.
A marketing team might say: increase sales of this product while maintaining a certain return. The advertising agent determines where and how to compete. On the other side, a shopping agent decides whether the resulting product deserves recommendation.
The important competition therefore shifts from individual ads toward the rules, data and interfaces through which machines negotiate commercial attention.
Mental Model: Machine-to-Machine Markets
Markets change when software can represent both buyers and sellers. Decisions that once required repeated human interaction become continuous automated negotiation.
Advertising stops being fully human-directed when machines begin buying attention from machines that allocate it.
Source: Amazon Ads
The System That Is Emerging
A new layer of commerce is forming between attention and transaction.
Call it the machine influence layer.
The old advertising system was designed around human perception. Brands bought exposure. Publishers sold audiences. Platforms measured clicks. Retailers monetized shelves.
AI introduces a different sequence.
A consumer expresses intent. An agent interprets it. Commercial information is gathered. Products become eligible or disappear. A shortlist is formed. Only then does the human encounter the result.
That creates new commercial assets: trusted machine-readable claims, inclusion inside agent conversations, access to recommendation systems, agent-readable publisher inventory and APIs through which advertising agents can act.
It also creates a difficult governance question. Where does useful commercial information end and paid manipulation begin?
A sponsored product on a webpage is visible. A sponsored signal consumed by an AI system may eventually influence an answer several steps removed from the original advertisement. The commercial relationship becomes harder for the customer to see.
That is why disclosure, provenance and rules around paid influence are likely to become infrastructure rather than cosmetic labels.
For CMOs, the strategic question is changing. It is no longer only: Where should we advertise?
Increasingly it becomes:
What information must machines trust before they are willing to recommend us?
Core Truth
The next advertising battle is not just for human attention. It is for legitimate influence over machine decisions.
