Criteo unveils data-driven AI commerce recommendation

Criteo has launched an Agentic Commerce Recommendation Service that connects AI shopping assistants with merchant inventory and returns curated product recommendations based on shopping and purchase signals.

The service targets a fast-growing segment of consumer technology, as large language model platforms add shopping features and retailers roll out chat-based product discovery tools. These systems are increasingly shaping how people search, compare and buy goods online.

Criteo is positioning the service as an alternative to approaches that rely mainly on product descriptions gathered from public sources. Instead, it draws on commerce data signals from real shopping activity, which can be difficult to capture through standard web crawling.

In internal testing, Criteo reported up to a 60% improvement in recommendation relevancy compared with third-party methods based only on product descriptions. It defines relevancy as how closely a product matches a shopper's intent, needs and preferences, and whether that match moves the shopper toward a purchase.

How it operates

The Agentic Commerce Recommendation Service is available through Criteo's Model Context Protocol and is designed to sit behind an AI assistant. A consumer makes a shopping request, and the assistant sends a query to Criteo. The system filters and ranks products using shopping and purchase signals, including popularity and availability, then returns a shortlist rather than a full catalogue.

The AI assistant presents and compares the recommended items and may support adding items to a basket or starting checkout within the same experience. The system can handle both broad, exploratory requests and product-specific searches, and can suggest complementary items when it determines the context is appropriate.

Criteo links performance to the scale of its commerce data. Its network covers 720 million daily shoppers, $1 trillion in annual transactions, and 4.5 billion product SKUs. Those signals, it argues, provide stronger context for what shoppers are likely to want than static listings alone.

The announcement comes as retailers and brands seek more control over how their products appear in AI-driven discovery channels. Recommendation logic is becoming a competitive battleground in digital commerce, as conversational interfaces and agent-style systems compress the steps between browsing and purchase.

Criteo's approach also reflects growing attention on data access in AI shopping. Product descriptions are widely available, but they can miss nuance around demand, conversion patterns, inventory status and other operational signals that influence what shoppers see and buy.

Market testing

Criteo has been testing the service with a major large language model platform since 2025 and is expanding tests with additional LLM platforms, retailers and brands.

For Criteo, the service extends efforts to embed its data and recommendation technology into emerging shopping journeys. The company has historically built its business around commerce media and performance advertising. Agent-led shopping could shift value from advertising placement toward recommendation infrastructure and transaction-linked decisioning.

Questions remain about how platforms will govern such integrations, including retailer permissions, brand presentation and user trust. Data privacy regulation and consumer expectations about the use of behavioural signals will also shape what models and intermediaries can use in practice.

Michael Komasinski, Criteo's chief executive officer, framed the launch as a data-driven contest between providers seeking to supply the recommendation layer for agentic shopping tools.

"The real competitive advantage in agentic commerce will come from access to high-quality commerce data at scale," said Michael Komasinski, Chief Executive Officer of Criteo. "This service brings that intelligence into AI-driven shopping experiences in a way that works for the entire ecosystem, delivering relevancy for consumers while respecting retailer data, brand integrity, and platform trust."

Criteo will continue testing the service with platform partners, retailers and brands as AI shopping assistants expand into mainstream consumer use.