Following hot on the heels of OpenAI launching Instant Checkout – a payment platform that allows ChatGPT users to buy straight from their AI interrogations – PayPal has integrated its wallet into the service. This is a massive boost for agentic AI, but has potentially profound implications for retail media.
The move is a vote of confidence in agentic commerce, with PayPal becoming the third major payment company to back this new way of shopping. Already in the market are Visa Intelligent Commerce and Mastercard Agent Pay, two bespoke agentic commerce and payment services run by each of the two major card companies. PayPal’s move to join forces with ChatGPT not only marks out the significance of agentic commerce, but also that ChatGPT is seen as a major platform to make it happen.
“Hundreds of millions of people turn to ChatGPT each week for help with everyday tasks, including finding products they love, and over 400 million use PayPal to shop,” says Alex Chriss, President and CEO of PayPal. “By partnering with OpenAI and adopting the Agentic Commerce Protocol, PayPal will power payments and commerce experiences that help people go from chat to checkout in just a few taps for our joint customer bases.”
And Chriss is not wrong. This agentic commerce – shopping powered by AI agents, where the consumer asks AI not only to find what they are looking for, but to actually autonomously go ahead and buy what AI surfaces – is generating significant hype as the future of online shopping. The total addressable market for agentic commerce has an estimated value of $136bn in 2025 and has been forecast to grow to hit a potential $1.7trn by 2030, according to Edgar, Dunn & Company.
Consumers too seem to be engaged. According to Cap Gemini, 71% of consumers want generative AI integrated into their shopping experiences; 68% want AI tools to aggregate results from search engines, social media and retailers. Further, a TechRadar-Omnisend survey in March 2025 found that 34% of US consumers would allow AI assistants to make purchases on their behalf.
How agentic commerce reframes retail media
While agentic commerce shifts how consumers search, shop online and buy online, the impact on how brands, products and services manifest on the internet in a way that agentic AI can find them reframes both what content goes online and how brands market their products. And this means a shift in retail media thinking.
Retail media today is about influencing human shoppers using first-party data within a retailer’s ecosystem. In agentic commerce, retail media becomes supplying the machine-readable signals and trust frameworks that AI agents need to decide and to recommend your product. So, the “audience” is shifting from human attention to AI interpretation.
For brands this means moving away from creative storytelling and impressive graphics, influencer marketing and special offers towards a world of structured data listings and playing along with the nascent world of AI influence ranking.
For example, a skincare brand today might run display ads on Boots or Target’s RMN. In an agentic world, its focus would shift to providing structured product metadata such as ingredients, efficacy and verified user data, while also offering transparent pricing and stock APIs. They will also have to optimizing to be “recommended” by personal AI agents when a user says, “Find me a gentle moisturizer under £30 with SPF.” So, brand advertising becomes machine-targeted influence – feeding data and trust signals rather than banner ads.
What changes for retailers with RMNs?
For retailers running retail media networks this is profound. They now must become data infrastructure providers for the agentic ecosystem, adding a raft of new ‘responsibilities’ to those running their first-party data and RMNs. This means exposing their product data and shopper signals in structured, AI-usable formats, such as Google’s structured product data or GS1 Digital Link, shifting retail media from “placement sales” to “data access and signal licensing.”
It also means offering agent-facing ‘ad inventory’, so instead of banner impressions, RMNs might sell priority response slots in AI shopping conversations. For example, when an AI assistant queries “best TV under €1,000,” an Elkjøp or MediaMarkt RMN could auction access to appear as the preferred offer in that recommendation. The onus will also be on retailers to provide authenticated reviews, return rates and ESG data that AI agents can trust.
These changes for retailers are profound, shifting their role from consumer data vendors to trust brokers. It will see them having to work with brands to co-train or feed LLMs with privacy-safe, commerce-specific data for mutual gain and see the transformation of data clean rooms into AI collaboration spaces.
It will also bring about a rethink in how retail media is measured; now agentic influence metrics, such as say the number of AI recommendations that lead to successful conversions will also now be a thing.
One truth, two outputs: how agentic works with traditional retail media
This is all well and good and will be a necessity as agentic commerce takes off. However, traditional banner and display ads and the other tropes of retail media aren’t going anywhere either any time soon. So how does a brand or retailer manage to service both?
Agentic commerce won’t replace human shopping overnight — it will layer on top of existing digital commerce. For the next decade, retailers and brands will operate in two modes simultaneously: traditional, human-targeted ecommerce and advertising and this agentic world. Both will need to coexist, because not all consumers will delegate decisions equally. Some will ask their agent to shortlist; others will still browse manually.
The good news is that the same underlying infrastructure can serve both worlds — provided it is modernised for machine readability. Agentic systems still depend on the same data backbone used for human-facing retail media: product titles, pricing, descriptions, reviews, images, stock, fulfilment, performance data and audience and behavioural analytics.
Think of it as one truth, two outputs; a single, enriched data layer feeding both human UX and AI interfaces. This means brands and retailers don’t need to double their work, but they must upgrade the structure and metadata of what they already manage.
In practice brands need to keep on as they are, but add new layers to their marketing processes. For example, they will need to continue writing compelling human-readable copy and maintain structured metadata with standardised attributes, verified claims, schema markup, sustainability scores and so on that AI agents can parse. On top of this, though, brands will have to adopt “content-for-machines” layers; PIM extensions that automatically translate marketing data into machine-readable schema.
For retailers this will mean building central data layers that power both retail media (ads) and AI interfaces (APIs). It will also see the adoption of a two-lane monetisation model across a traditional lane of sponsored placements, audience targeting, brand storytelling et al, and an agentic lane for paid access to verified data, priority listings in agent responses and algorithmic bidding for agent recommendations.
Interestingly, AI itself will be used to auto-generate and maintain structured data. For example, a retailer uses AI to continuously check brand listings for missing attributes or inconsistent claims, ensuring both human pages and agent feeds are accurate.
Finally, as said, metrics will also have to be redefined for retail media, with a move from “impressions” to “influence signals”, both being tracked in parallel until one dominates.
You are in: Home » Retail Media » ANALYSIS Parallel universe: how Agentic Commerce reframes retail media
ANALYSIS Parallel universe: how Agentic Commerce reframes retail media
Paul Skeldon
Following hot on the heels of OpenAI launching Instant Checkout – a payment platform that allows ChatGPT users to buy straight from their AI interrogations – PayPal has integrated its wallet into the service. This is a massive boost for agentic AI, but has potentially profound implications for retail media.
The move is a vote of confidence in agentic commerce, with PayPal becoming the third major payment company to back this new way of shopping. Already in the market are Visa Intelligent Commerce and Mastercard Agent Pay, two bespoke agentic commerce and payment services run by each of the two major card companies. PayPal’s move to join forces with ChatGPT not only marks out the significance of agentic commerce, but also that ChatGPT is seen as a major platform to make it happen.
“Hundreds of millions of people turn to ChatGPT each week for help with everyday tasks, including finding products they love, and over 400 million use PayPal to shop,” says Alex Chriss, President and CEO of PayPal. “By partnering with OpenAI and adopting the Agentic Commerce Protocol, PayPal will power payments and commerce experiences that help people go from chat to checkout in just a few taps for our joint customer bases.”
And Chriss is not wrong. This agentic commerce – shopping powered by AI agents, where the consumer asks AI not only to find what they are looking for, but to actually autonomously go ahead and buy what AI surfaces – is generating significant hype as the future of online shopping. The total addressable market for agentic commerce has an estimated value of $136bn in 2025 and has been forecast to grow to hit a potential $1.7trn by 2030, according to Edgar, Dunn & Company.
Consumers too seem to be engaged. According to Cap Gemini, 71% of consumers want generative AI integrated into their shopping experiences; 68% want AI tools to aggregate results from search engines, social media and retailers. Further, a TechRadar-Omnisend survey in March 2025 found that 34% of US consumers would allow AI assistants to make purchases on their behalf.
How agentic commerce reframes retail media
While agentic commerce shifts how consumers search, shop online and buy online, the impact on how brands, products and services manifest on the internet in a way that agentic AI can find them reframes both what content goes online and how brands market their products. And this means a shift in retail media thinking.
Retail media today is about influencing human shoppers using first-party data within a retailer’s ecosystem. In agentic commerce, retail media becomes supplying the machine-readable signals and trust frameworks that AI agents need to decide and to recommend your product. So, the “audience” is shifting from human attention to AI interpretation.
For brands this means moving away from creative storytelling and impressive graphics, influencer marketing and special offers towards a world of structured data listings and playing along with the nascent world of AI influence ranking.
For example, a skincare brand today might run display ads on Boots or Target’s RMN. In an agentic world, its focus would shift to providing structured product metadata such as ingredients, efficacy and verified user data, while also offering transparent pricing and stock APIs. They will also have to optimizing to be “recommended” by personal AI agents when a user says, “Find me a gentle moisturizer under £30 with SPF.” So, brand advertising becomes machine-targeted influence – feeding data and trust signals rather than banner ads.
What changes for retailers with RMNs?
For retailers running retail media networks this is profound. They now must become data infrastructure providers for the agentic ecosystem, adding a raft of new ‘responsibilities’ to those running their first-party data and RMNs. This means exposing their product data and shopper signals in structured, AI-usable formats, such as Google’s structured product data or GS1 Digital Link, shifting retail media from “placement sales” to “data access and signal licensing.”
It also means offering agent-facing ‘ad inventory’, so instead of banner impressions, RMNs might sell priority response slots in AI shopping conversations. For example, when an AI assistant queries “best TV under €1,000,” an Elkjøp or MediaMarkt RMN could auction access to appear as the preferred offer in that recommendation. The onus will also be on retailers to provide authenticated reviews, return rates and ESG data that AI agents can trust.
These changes for retailers are profound, shifting their role from consumer data vendors to trust brokers. It will see them having to work with brands to co-train or feed LLMs with privacy-safe, commerce-specific data for mutual gain and see the transformation of data clean rooms into AI collaboration spaces.
It will also bring about a rethink in how retail media is measured; now agentic influence metrics, such as say the number of AI recommendations that lead to successful conversions will also now be a thing.
One truth, two outputs: how agentic works with traditional retail media
This is all well and good and will be a necessity as agentic commerce takes off. However, traditional banner and display ads and the other tropes of retail media aren’t going anywhere either any time soon. So how does a brand or retailer manage to service both?
Agentic commerce won’t replace human shopping overnight — it will layer on top of existing digital commerce. For the next decade, retailers and brands will operate in two modes simultaneously: traditional, human-targeted ecommerce and advertising and this agentic world. Both will need to coexist, because not all consumers will delegate decisions equally. Some will ask their agent to shortlist; others will still browse manually.
The good news is that the same underlying infrastructure can serve both worlds — provided it is modernised for machine readability. Agentic systems still depend on the same data backbone used for human-facing retail media: product titles, pricing, descriptions, reviews, images, stock, fulfilment, performance data and audience and behavioural analytics.
Think of it as one truth, two outputs; a single, enriched data layer feeding both human UX and AI interfaces. This means brands and retailers don’t need to double their work, but they must upgrade the structure and metadata of what they already manage.
In practice brands need to keep on as they are, but add new layers to their marketing processes. For example, they will need to continue writing compelling human-readable copy and maintain structured metadata with standardised attributes, verified claims, schema markup, sustainability scores and so on that AI agents can parse. On top of this, though, brands will have to adopt “content-for-machines” layers; PIM extensions that automatically translate marketing data into machine-readable schema.
For retailers this will mean building central data layers that power both retail media (ads) and AI interfaces (APIs). It will also see the adoption of a two-lane monetisation model across a traditional lane of sponsored placements, audience targeting, brand storytelling et al, and an agentic lane for paid access to verified data, priority listings in agent responses and algorithmic bidding for agent recommendations.
Interestingly, AI itself will be used to auto-generate and maintain structured data. For example, a retailer uses AI to continuously check brand listings for missing attributes or inconsistent claims, ensuring both human pages and agent feeds are accurate.
Finally, as said, metrics will also have to be redefined for retail media, with a move from “impressions” to “influence signals”, both being tracked in parallel until one dominates.
Read More
You may also like
Subscribe to our email community