At CES in January, I witnessed the pace of AI development across media and advertising on full display, with capabilities that were previously being piloted moving towards deployment as core infrastructure within organisations.
Just ten weeks later, as I prepare to join industry leaders at Advertising Week Europe, that momentum has only intensified. Three announcements since CES signal where the market is heading, and how intentionally enterprises need to move to keep pace.
Agentic commerce becomes infrastructure
Google and Shopify began 2026 with the launch of the Universal Commerce Protocol. UCP is an open standard for agentic commerce co-endorsed by Walmart, Target, Visa, Mastercard, Stripe, and over 20 industry partners, designed to establish a shared language through which AI agents can discover, evaluate, and complete transactions across any conversational surface, without bespoke integrations for each platform.
As we explored in our analysis of the announcement with Michael Dobell, Co-founder of Monks and Vice-Chair of SoDA, what distinguishes UCP is its merchant-centric architecture. Where competing approaches prioritise frictionless customer experience – sometimes at the cost of merchant control – UCP’s layered design allows retailers to adopt incrementally, opting into capabilities that match their operational maturity.
But as that conversation also surfaced, infrastructure adoption and user adoption operate on different timelines. And critically, the architectural decisions being made now – which protocol layers to adopt, how to integrate with existing commerce stacks, how to govern AI-executed transactions – will determine who is positioned to act effectively when consumer behaviour catches up.
ChatGPT becomes an advertising platform
Also in January, OpenAI confirmed that advertisements will begin appearing in ChatGPT. In early March, Criteo became OpenAI’s first advertising technology partner, with the integration now rolling out across ChatGPT’s free and Go tiers. For advertisers and agencies, this represents more than an advertising surface: it introduces a conversational context where established targeting, measurement, and attribution models don’t map cleanly onto how users interact with AI-generated responses.
OpenAI’s approach departs from established advertising models in several ways. Targeting is contextual instead of behavioural, as ads are matched to the topic of a live conversation rather than a tracked user profile. Success is measured not by click conversions but by becoming the recommended solution within the response itself. And OpenAI is taking a restrictive approach to access, with reported CPMs around $60 – roughly triple Meta’s average and on par with premium CTV – limiting entry to major brands.
Advertisers are largely trusting OpenAI’s internal reporting for now, a dynamic the industry hasn’t comfortably accepted since the early days of social platforms. The lack of independent measurement and attribution infrastructure creates operational challenges for buyers managing cross-platform campaigns where other channels offer established verification standards.
AI agents become native to advertising platforms
Then, in late February, Meta embedded Manus AI – the autonomous agent technology it acquired in a notable $2 billion transaction in late 2025 – directly into its Ads Manager interface that millions of advertisers use daily.
Manus now sits natively in the platform’s navigation, designed to execute multistep tasks autonomously: conducting market research, generating performance reports, analysing audience segments, and optimising campaigns based on advertiser instructions. The integration means autonomous campaign management is no longer an adjacent capability offered by specialist vendors; it’s becoming embedded infrastructure within the platforms where budget decisions are made.
When one of the industry’s largest advertising platforms embeds autonomous agents as core infrastructure rather than optional tooling, it resets expectations industry-wide. And for teams of media buyers and agencies operating at enterprise scale, developing fluency with autonomous workflows has become a strategic priority and an operational requirement.
The competitive landscape, beyond technical capability
The velocity of these announcements is matched by the intensity of competition as platforms stake opposing positions on how AI should integrate into commercial systems.
Google and Shopify’s UCP launched to challenge OpenAI’s Instant Checkout, as both vie to become the standard for AI-mediated transactions. While UCP positions itself as an open standard backed by major retailers and payment providers, OpenAI’s approach keeps transactions within its own ecosystem.
Separately, Anthropic and OpenAI have taken opposite stances on monetisation. Anthropic’s Super Bowl spot this year ran with the tagline “Ads are coming to AI. But not to Claude.”, framing ad-free as a trust and neutrality feature for enterprise clients concerned about commercial bias in AI recommendations, in direct contrast to OpenAI’s decision to launch advertising in ChatGPT at scale.
That positioning became consequential last week when Anthropic’s refusal to provide unrestricted Pentagon access, drawing firm lines against autonomous weapons and mass surveillance, led to a retaliatory supply chain risk designation. Regulated-sector brands, particularly in financial services and defence, are now reassessing whether they can continue using Anthropic’s platforms without triggering compliance issues. The dispute makes explicit that platform choice carries strategic, brand, and regulatory implications, with competition increasingly playing out through political and ethical positioning as much as technical capability.
Why enterprise readiness matters more than opportunity
These three announcements, and the competitive dynamics they surface, together are reshaping the entire advertising and commerce stack – across transaction protocols, advertising surfaces, and campaign execution mechanisms.
Together, they illustrate how quickly the transactional foundations of advertising and commerce are being rebuilt, and signal a market moving with velocity that enterprises must match. It is the architectural decisions enterprises make now that will determine whether they are positioned to capitalise on this shift or are left adapting to a market that has already moved. Organisations now need orchestration layers capable of governing multiple AI agents, protocols, and advertising environments simultaneously, and the ability to maintain measurement integrity when AI mediates both the buy and the optimisation.
As the industry convenes at Advertising Week Europe, the conversation will naturally focus on opportunity and the next wave of advancements that will follow. But in my conversations with media organisations, agencies, and platforms, the more substantive discussions I expect to have will focus on operational readiness and whether they have the internal foundations needed to adopt these capabilities without compromising future control or competitive position.
