LiveRamp today announced a set of agentic AI pilots designed to help commerce media networks move faster from consumer data to measurable campaign outcomes, targeting an operational problem that has slowed advertiser adoption across food delivery, big box retail, and grocery segments.

The announcement, published on the LiveRamp blog on June 21, 2026, describes three early pilot patterns tied to specific business goals. According to LiveRamp, the underlying premise is that commerce media networks hold differentiated assets - first-party transaction data, direct consumer relationships, and media inventory close to the point of purchase - but face persistent difficulty turning those assets into repeatable, scalable advertiser outcomes.

The problem LiveRamp is trying to solve

Commerce media networks occupy a structurally advantageous position in the advertising ecosystem. Whether the network belongs to a grocery chain, a food delivery platform, an airline, or a financial institution, the core data proposition is the same. According to LiveRamp, "the network knows things about its consumers that no one else can replicate." The challenge, according to the company, is not access to that data. It is operationalizing that data into repeatable, scalable outcomes for multiple advertisers simultaneously, through workflows that are governed, fast, and reproducible.

What has made this difficult in practice is a set of operational bottlenecks that LiveRamp describes as "fragmented workflows, one-off analysis, and slow time to value." Social platforms and demand-side platforms (DSPs) have spent years simplifying advertiser experience: a marketer sets a budget, defines a goal, and lets automated systems handle the optimization. Advertisers have grown accustomed to that model. They are increasingly asking commerce media networks to offer something comparable. Most CMNs, despite holding richer first-party data than any social platform, cannot yet deliver that level of automation.

That gap is the direct target of the pilots. LiveRamp frames agentic AI as the mechanism for closing it - reducing the manual steps that slow down campaign teams and enabling more time for strategic work rather than operational assembly.

How the pilots are structured

The agentic AI approach LiveRamp is building does not start from scratch. According to the company, its approach "builds on the collaboration infrastructure CMNs and advertisers already use today, securely layering agentic orchestration on top of existing clean room environments, APIs, and interfaces rather than asking teams to start from scratch." That framing positions the pilots as extensions of infrastructure already in use rather than replacements for it.

The LiveRamp platform is described as "fundamentally designed for neutrality and interoperability," enabling customers to build their own agents, leverage partner agents, or use what LiveRamp calls the LiveRamp Agentic Builder program. The architecture is designed to accommodate multiple approaches rather than locking customers into a single proprietary agent stack.

Three distinct pilot patterns are currently active, each associated with a different commercial context.

Food delivery pilots

CMNs built on transaction and behavioral data from food delivery services are using the agentic approach to pursue net-new customer acquisition, higher purchase frequency, and customer reactivation. Food delivery platforms sit on particularly dense behavioral signals - what consumers order, how often, at what time, and from which merchants - and the pilot is designed to convert those signals into automated campaign workflows without requiring manual analysis cycles between each step.

Big box and specialty retail pilots

Retailers with broad product assortments are using first-party intelligence to support product launches, expand basket size, and deepen customer loyalty through "more targeted, timely engagement," according to LiveRamp. The challenge for large-format retailers is that the volume of products, advertiser relationships, and audience segments makes manual campaign management operationally intensive. Agentic orchestration, in this context, is primarily a throughput problem.

Grocery retail pilots

Grocery CMNs are using the pilots to gain "stronger performance visibility and smarter cross-channel optimization," with clearer insight into reach, frequency, conversions, incrementality, and ROAS (return on ad spend), according to LiveRamp. Grocery is the most established vertical in retail media investment, with major networks including Albertsons Media Collective and Kroger Precision Marketing having built out significant measurement infrastructure. The pilots in this segment focus specifically on performance visibility and cross-channel optimization rather than acquisition or loyalty mechanics.

What is still being built

LiveRamp is transparent about the areas where the system is incomplete. According to the company, four dimensions of the agentic workflow are still being actively developed.

Governance and permissions remain partly unresolved at the industry level. LiveRamp states that while industry-wide standards for agent permissions and data controls are still being developed, it already provides platform controls that allow retailers and brands to define what is permitted, who can access data, and which workflows are allowed, with those rules enforced in the system. That foundation, according to LiveRamp, is what makes more advanced agentic workflows possible while respecting each party's privacy requirements.

User experience design for outcome-first agent workflows in media is also still taking shape. LiveRamp states the company believes "the future is a workflow-native planning experience, not another chatbot." But the specific design patterns for how practitioners interact with those agents in day-to-day planning are still being refined based on practitioner feedback.

Activation handoffs - the moment when an agent transitions from insight generation to actual campaign activation - need to feel seamless rather than requiring manual intervention between steps. According to LiveRamp, the exact shape of that handoff, including what APIs return, how projected match rates are displayed, and how alternative targeting options are presented, is still being refined with customers and partners.

Measurement and optimization loops represent the most technically complex outstanding area. An outcome-first agent needs to do more than plan and activate campaigns. It also needs to learn from measurement results and use those insights to inform future decisions. According to LiveRamp, that closed-loop system - where campaign performance data feeds back into agent decision-making - is a work in progress and remains an active area of development.

This level of specificity about what is not yet finished is unusual in product announcements. The company is explicitly signaling that it is building this capability in collaboration with a select group of customers rather than releasing a finished product.

Context: LiveRamp's agentic AI trajectory

The commerce media pilots sit within a broader and accelerating pattern of agentic AI development at LiveRamp. The company launched its first agentic AI orchestration capabilities on October 1, 2025, positioning itself as the first data collaboration platform to give AI agents governed access to its complete marketing tool suite, spanning identity resolution, segmentation, activation, measurement, clean rooms, and a partner network of over 900 brands, publishers, and platforms.

On November 3, 2025, LiveRamp donated its User Context Protocol to the IAB Tech Lab, establishing an open standard for how AI agents exchange identity, contextual, and reinforcement signals across advertising systems. The protocol uses dense vector representations of 256 to 1,024 dimensions to compress signals into formats fast enough for real-time inference.

In January 2026, the company expanded its Data Marketplace to include licensing infrastructure for AI training datasets, pre-built AI models, and AI-powered applications. That expansion transformed what had been principally an audience marketplace into broader AI infrastructure. Two live agent integrations followed on March 3, 2026: Newton Research agents for cross-media measurement and SemantIQ agents for healthcare provider audience building, both operating directly within the LiveRamp platform.

April 2026 brought two more additions. LiveRamp partnered with Akkio to integrate conversational AI into measurement reports. Separately, the company added NVIDIA GPU infrastructure to its clean rooms, enabling AI model training and inference at up to 15 times the speed of CPU-based environments - directly addressing compute constraints that had limited the practical capability of agentic workloads running against large datasets.

On June 17, 2026 - five days before today's commerce media announcement - LiveRamp launched the LiveRamp Agent Builders (LAB) program, a formal initiative allowing third-party AI companies to bring purpose-built agents onto its platform and make them available to LiveRamp's full customer base via APIs and MCP servers. The four founding partners were SemantIQ, Newton Research, Akkio, and Datalinx.

Taken sequentially, the commerce media pilots announced today are the domain-specific application layer sitting on top of that infrastructure stack: orchestration framework (October 2025), open protocol standard (November 2025), expanded data availability (January 2026), live agent integrations (March 2026), compute capacity upgrade (April 2026), third-party builder program (June 17, 2026), and now vertical-specific pilots for commerce media (June 22, 2026).

The retail media context

The timing of the commerce media pilots connects to a broader shift in how the retail media sector is developing. LiveRamp expanded its retail media measurement capabilities to include Meta campaign attribution through its Clean Room platform on October 23, 2025, enabling networks such as Albertsons Media Collective and Roundel (Target's retail media arm) to connect Meta advertising exposure with first-party sales data inside a privacy-safe environment.

That work demonstrated a structural point relevant to the agentic pilots: the same clean room infrastructure that enables post-campaign measurement can, in principle, serve as the foundation for automated campaign planning and optimization. The Albertsons and Roundel implementations showed that retailers already trust LiveRamp's infrastructure for sensitive data matching. The agentic layer is asking those same retailers to extend that trust to automated decision-making within the same environment.

The DoorDash Ads expansion announced on June 4, 2026 is directly relevant to the food delivery pilot pattern. DoorDash announced a LiveRamp clean room partnership at that time, with early data showing that over 80% of consumers reached through the integration were new to advertisers - a net-new customer acquisition signal that mirrors one of the explicit goals LiveRamp describes for its food delivery CMN pilots. The data matching mechanism works through the same infrastructure: advertiser first-party data and DoorDash first-party data are matched inside LiveRamp's clean room environment, allowing overlap and incrementality analysis without either party accessing the other's raw data.

The grocery retail pilot pattern also connects to visible industry activity. Grocery TV added third-party sales lift measurement via ABCS Insights in April 2026, using a panel of 41 million US households to provide independent, purchase-verified attribution to brands advertising on its in-store network. The measurement infrastructure challenges that Grocery TV's work addresses - the need for incrementality measurement, conversion tracking, and cross-channel ROAS visibility - are precisely the outcomes that LiveRamp's grocery CMN pilots target.

Why this matters for marketers

The advertising technology industry has been building toward agentic campaign management across multiple fronts simultaneously. The IAB's 2026 Outlook Study forecast 9.5% US advertising spend growth, with two-thirds of advertisers now focusing on agentic AI for campaign execution. Commerce media specifically was projected to grow 12.1% in that forecast. Yet the same research and industry observers have consistently noted that most organizations lack the unified infrastructure to deploy agentic systems effectively.

The commerce media problem that LiveRamp is addressing is specific and commercially meaningful. CMNs have first-party data that is genuinely scarce - actual transaction records from grocery purchases, food delivery orders, or big-box retail visits - but translating that data into advertiser value requires analytical workflows, audience segmentation, campaign planning, activation, and measurement steps that currently require significant human coordination. Each step introduces delay, and the aggregate delay is what makes CMN advertising slower and more operationally intensive than comparable buys on social platforms or major DSPs.

According to LiveRamp, "the CMNs that standardize and accelerate the path from consumer insight to campaign to measured outcome will be the ones that win advertiser investment." The statement describes a competitive dynamic that is already visible in the market. Retail media networks that can offer automation comparable to social platforms are better positioned to attract brand budgets that would otherwise default to established platforms.

Publicis Groupe's pending $2.5 billion acquisition of LiveRamp, announced on May 18, 2026 at a 30% premium to LiveRamp's prior trading price and pending regulatory approvals and a shareholder vote, adds a further dimension to the commerce media pilots. Publicis described a retailer building a "retail journey agent across CRM, loyalty, in-store, and retail media inventory" as one of the central use cases for the combined entity. The agentic retail media workflow that LiveRamp is now piloting with select CMNs is, in structural terms, precisely that use case. The pilots effectively constitute a proving ground for what the combined Publicis-LiveRamp entity may eventually offer at scale.

Timeline

Summary

Who: LiveRamp (NYSE: RAMP), a San Francisco-based data collaboration platform currently subject to a pending $2.5 billion acquisition by Publicis Groupe, announced the pilots in collaboration with a select group of commerce media network customers and ecosystem partners across three verticals.

What: Three early agentic AI pilot patterns for commerce media networks, layering automated orchestration on top of existing clean room environments, APIs, and interfaces to connect consumer insight to campaign activation, measurement, and optimization. Pilots span food delivery (targeting net-new customer acquisition, purchase frequency, and reactivation), big box and specialty retail (product launches, basket expansion, and loyalty), and grocery (cross-channel ROAS, reach, frequency, and incrementality visibility).

When: The LiveRamp blog post describing the pilots was published on June 21, 2026. The public announcement followed on June 22, 2026.

Where: The pilots operate on LiveRamp's existing data collaboration infrastructure, including its Clean Room environments, APIs, and partner integrations. Commerce media networks interested in participation are directed to LiveRampAgenticPilots@liveramp.com.

Why: Commerce media networks hold scarce first-party transaction data but face an operational gap in converting that data into repeatable, scalable advertiser outcomes. Fragmented workflows, one-off analysis, and slow time-to-value make CMN advertising operationally slower than social and DSP alternatives, putting CMNs at a competitive disadvantage for advertiser budget allocation. LiveRamp is positioning governed agentic AI - layered on top of existing clean room and identity infrastructure - as the mechanism for closing that operational gap without requiring CMNs or brands to rebuild existing workflows from scratch.