Pacvue Prism bets on agentic AI to unify the fractured commerce stack
Pacvue, the commerce media platform that the company says handles more than $12 billion in annual advertising spend, has launched Pacvue Prism, a product designed to collapse the fragmented architecture of modern brand marketing into what it calls an Agentic Commerce Grid. The launch, announced at Cannes Lions on 22 June, positions agentic AI not as a creative tool but as an operational layer that routes media budgets toward business outcomes rather than channel-specific metrics.
The timing is deliberate. As generative AI rewires how consumers discover products, researching through large language models, watching shoppable video, and completing purchase through a retailer's app, sometimes in a single session, the legacy model of siloed channel teams, each with their own budget and measurement dashboard, is increasingly incoherent. Pacvue's argument is that no existing platform was built for this non-linear journey, and Prism is designed to be the first that is.
From channel optimisation to commerce accountability
The platform's core claim is measurement coherence. Prism connects every media touchpoint, from CTV and programmatic display to conversational search and shoppable video, to underlying commerce signals: actual sales, new-to-brand customer acquisition, and revenue incrementality. Attribution is handled through a proprietary regression modelling framework paired with pixel tracking, designed to give CMOs a single answer to the question of what a given dollar of media spend is actually driving.
Alpha-stage participants have reported early figures from the platform: a 34% increase in impressions, a 23% reduction in cost per thousand (CPM), a 12% lift in total sales, and 24% growth in new-to-brand customers. Pacvue acknowledges these are early-stage results from a controlled cohort. "Commerce media is moving from channel optimisation to commerce accountability," said Rahul Choraria, CEO of Pacvue. "The brands that win from here will operate at the grid level: every touchpoint connected, every dollar accountable."
The agentic dimension of the platform goes beyond automated bidding. Prism is built to connect into a brand's broader data infrastructure, including data lakes, customer data platforms, CRMs, and custom AI agents via the Model Context Protocol (MCP), a technical standard for agent-to-agent communication that has gained traction across enterprise AI stacks in 2025-26. That integration means performance signals from one channel theoretically sharpen decision-making across all others over time.
The convergence angle: AI infrastructure meets the $200bn retail media market
Retail media has grown into one of the highest-margin advertising categories globally, with eMarketer and other research firms projecting the market to exceed $200 billion by the late 2020s. That growth has attracted both endemic commerce platforms and general-purpose advertising technology players, a structural tension the Prism launch directly addresses by positioning Pacvue as the orchestration layer above both.
The more consequential convergence signal, however, is architectural. Pacvue Prism is essentially a bet that the infrastructure logic of agentic AI, autonomous agents acting on objectives rather than instructions, will migrate from enterprise software into marketing operations at scale. If that transition accelerates, it has implications well beyond ad tech. Brand and agency capital that currently funds human planning and channel management teams could shift toward AI infrastructure licensing, reshaping the economics of agencies, specialist media buyers, and the broader martech vendor landscape.
For cross-sector investors, the relevant read-across is to enterprise software: the same agentic-layer thesis driving valuations in sales automation and supply chain AI is now arriving in commerce media, a category with measurable, high-frequency outcome data. That feedback loop, real transaction data training the agents optimising the spend that generates more transaction data, is structurally more defensible than agentic applications in domains where ground truth is harder to define. Pacvue's reported 12% share of global retail media spend, excluding China, gives it a data-volume advantage that will be difficult for newer entrants to replicate quickly.