Users have already started asking AI agents for product recommendations. A fundamental question is: when ChatGPT recommends a product, who pays whom? When an AI agent books a flight for you, how does the airline compensate the platform that surfaced the product? When Perplexity cites a retailer in its answer, is there a commercial arrangement behind that?
The answers to these questions form the monetization layer of agentic commerce. It is a core engine that turns AI recommendations into real money, and it is evolving fast in 2026.
Let’s break down how AI agents get paid, why it matters for merchants, and where the ecosystem is heading.
Why This Question Suddenly Matters
In the past several years, AI chatbots did not really monetize commerce. They answered questions and pointed users to a merchant website. Users clicked through, and the merchant handled everything from there.
That changed quickly. ChatGPT alone processes roughly 50 million shopping queries per day. AI-referred retail traffic is up 393% year-over-year.
As AI becomes central to commerce, we need to understand how the money flows. That is exactly what is happening—often behind the scenes, as new networks and protocols form.
The Four Ways AI Agents Make Money
Right now, AI agent monetization falls into four broad models:
1. Subscription and Usage Fees
Some AI agents charge users directly through subscription models like ChatGPT Plus or usage-based models like API pricing. This is the most popular approach. Revenue here does not come from merchants; it comes from users who benefit directly from AI agents.
2. Advertising Revenue
Advertising revenue is the traditional model adapted for conversational AI. When an AI assistant surfaces a product or service, the merchant pays for that placement. OpenAI began rolling this out to free ChatGPT users in May 2026 and reached a $1 billion ARR run rate in August. Google’s AI Overviews now include commercial placements, and Perplexity has sponsored answers.
The pricing structures vary. Some charge on a CPC, or cost per click, basis; others charge CPA, or cost per action, or CPS, or cost per sale. Outcome-based models are increasingly attractive because they align advertiser risk with actual results.
3. Affiliate Commissions
When an AI agent recommends a product and a user converts—for example, by placing an order—the platform earns a commission. This is a natural fit for AI agents because it relates revenue directly to real purchase intent rather than page views or impressions.
Traditional affiliate networks such as Amazon Associates, Awin, and Impact have started adapting to serve AI agents. But there is a fundamental mismatch—as one industry analysis shows—because almost every affiliate tool ever built assumes a static webpage, not a streamed LLM response.
New infrastructure is emerging to handle this, including networks specifically built to connect AI agents to merchant offers.
4. Direct Transactions
Some AI agents do not just recommend products—they complete the loop by supporting checkout inside the AI experience. Here, revenue comes from a share of the transaction itself, similar to how marketplaces such as Uber or Airbnb work.
Shopify’s Agentic Storefronts, launched March 24, 2026, gave 5.6 million stores access to ChatGPT and Gemini with built-in attribution. This is the kind of infrastructure that lets AI agents complete transactions more cleanly.
Hybrid models are also appearing: a base subscription for usage, combined with advertising fees or commissions on transactions that an agent facilitates.
The Infrastructure Layer That Makes It Work
Agentic commerce does not just happen inside ChatGPT. An entire technical infrastructure is forming to make it work.
When an AI agent recommends a product to a user, it needs:
- A structured way to discover available merchant offers.
- Verification that the offer is real, in stock, and priced correctly.
- Attribution tracking so the platform knows when conversions follow.
- Settlement—the mechanism that actually pays the platform when a transaction happens.
That is not something an AI agent handles alone. It needs infrastructure, networks, and protocols.
This is where platforms such as Agent Offer Network (AON) come in. AON is a structured offer layer that connects AI apps and agents to merchant offers, handling discovery, attribution, and settlement behind the scenes. Merchants publish offers into the network, AI agents query and surface matched offers, and attribution and payment can then flow through the system.
You do not need to understand every technical detail as a merchant. But knowing this layer exists helps you understand how AI agents surface purchasable products—not just generic descriptions.
The Attribution Problem and Why It Matters
Here is a challenge that runs through the entire monetization layer: attribution in agentic commerce is genuinely hard.
Traditional tracking relies on cookies and pixels. Those methods can fail when a conversation happens inside a chat interface and the eventual purchase happens on a completely different platform. A user might chat with an AI agent on their phone, then buy on their laptop three days later. Was it an AI-driven purchase? Traditional analytics might label it as direct traffic.
This is why infrastructure networks matter. They can provide server-side attribution—tracking that does not depend on cookies or the same-device journey. When attribution is clean, outcome-based pricing such as CPA and CPS becomes possible. Otherwise, systems may fall back to less efficient models such as CPC or CPM.
Full attribution maturity in agentic commerce is still estimated to be 18–24 months away. The article presents this as an external estimate, not a guaranteed timeline. Merchants that invest in the infrastructure early can build useful data before the framework becomes more mature.
What This Means for Merchants
We can now see a clearer picture when these developments are considered together:
- Agentic commerce is becoming important infrastructure. The money is already flowing. If your business is not connected to the layer that connects AI agents to merchant offers, you may miss an increasingly important source of demand.
- Outcome-based pricing is increasingly important. Platforms are willing to price on results because AI-driven traffic can be tied more closely to measurable actions.
- Attribution matters more than ever. Merchants that invest in server-side tracking and clean attribution can make AI-assisted ROI easier to measure.
- Networks may consolidate. The agentic commerce infrastructure layer is currently fragmented across many networks and protocols. The article expects consolidation over the next 24 months, although the timing remains uncertain.
The Bigger Picture
Agentic commerce is going through a phase similar to the one programmatic advertising went through 15 years ago: messy, fragmented, and complex, but forming into real infrastructure quickly. The networks that emerge from this phase may become the plumbing that powers billions of dollars in AI-driven commerce.
Right now, most merchants are not paying attention to this new layer. They are still focused on their own websites and traditional funnels. But the reality is that the agentic commerce ecosystem being built underneath them may influence which businesses AI agents recommend over the next decade.
The merchants who understand this early can gain an advantage—not by building the infrastructure themselves, but by making their businesses easy to connect to it.
FAQ
What does the monetization layer of agentic commerce actually do?
It connects AI-assisted product discovery to a measurable commercial outcome. The layer supports structured offer discovery, verification of details such as price and availability, attribution of clicks or conversions to an AI surface, and settlement when an eligible event occurs. The article uses Agent Offer Network (AON) as an example of this type of structured offer layer. For merchants, the practical value is that an AI recommendation can become measurable demand rather than an untracked referral.
What should merchants know about the four ways AI agents monetize commerce?
AI agents generally monetize commerce through four models: user-paid subscriptions or usage fees; merchant-paid advertising; affiliate commissions on referred conversions; and revenue from direct transactions completed inside the agent experience. The article points to commercial placements in ChatGPT, Google AI Overviews, and Perplexity, and uses Shopify’s Agentic Storefronts as an example of direct transaction infrastructure. It also cites rapid demand growth, including roughly 50 million ChatGPT shopping queries per day and Adobe’s reported 393% year-over-year increase in AI-referred retail traffic. For merchants, the model determines whether they pay for exposure, action, or sale.
Which downstream events can make an AI-assisted interaction payable?
A defined downstream event—not the recommendation alone—can make an AI-assisted interaction potentially payable. Depending on the commercial model, that event may be a qualified click, lead, booking, app action, or confirmed purchase. The merchant and platform should agree on the event definition, validation rules, attribution window, and settlement terms before evaluating performance. A click may support CPC pricing, while CPA and CPS require a meaningful action or completed sale.
How can I measure an AI-assisted purchase when the customer converts on another device?
Use attribution that can connect the AI interaction, offer or click identifier, and downstream conversion across sessions or devices. Cookie- and pixel-based tracking can miss a journey in which a customer chats with an AI agent on a phone and purchases from a laptop days later. Server-side attribution can provide a stronger connection, but you should still separate AI-assisted outcomes from direct traffic and define which events qualify for settlement.
How mature is attribution for agentic commerce today?
It is still developing and is not fully standardized. The article cites an external analysis estimating that full attribution maturity may still be 18–24 months away; treat that as a directional forecast, not a guaranteed timeline. In the meantime, merchants should prioritize reliable conversion tracking, define which events qualify for payment, and use server-side or cross-session attribution where available.
Will agentic commerce networks consolidate, and what should merchants do now?
Possibly, but the timing is uncertain. The article compares agentic commerce with the early development of programmatic advertising and presents network consolidation over the next 24 months as an expectation, not a guaranteed schedule. A practical response is to keep offer data, conversion definitions, attribution records, and settlement terms clearly documented so you can review partners as the ecosystem and its protocols evolve.
Do I need to build my own agentic commerce infrastructure to participate?
No. You generally do not need to build the entire discovery, attribution, and settlement stack yourself. A network or protocol can connect structured merchant offers to AI applications, while your team maintains accurate product data, pricing, availability, policies, and conversion signals. Before relying on a partner, confirm how it handles offer updates, attribution, settlement, and disclosure for sponsored placements.
