Singapore – Criteo has introduced its ‘Agentic Commerce Recommendation Service,’ a new solution designed to power AI shopping assistants with accurate, commerce-grade product recommendations built on real-world shopping data.
As large language models evolve into AI shopping assistants and retailers deploy their own conversational tools, the way consumers discover and purchase products online is rapidly changing.
Criteo’s new service addresses the need for recommendation infrastructure that goes beyond publicly available product descriptions by drawing on real shopping behaviour to deliver trusted and personalised results.
Built on Criteo’s commerce intelligence, the service demonstrated up to a 60% improvement in recommendation relevancy compared with third-party approaches based solely on product descriptions, according to internal testing conducted in January 2026.
The performance is supported by Criteo’s scale, which includes 720 million daily shoppers, $1 trillion in annual transactions, and 4.5 billion product SKUs.
Available through Criteo’s Model Context Protocol, the service connects AI-powered shopping assistants directly with merchant inventory, translating consumer queries into curated, transaction-ready product recommendations.
By applying real-world shopping and purchase signals, it enables AI assistants to surface products that best match individual intent, availability, and popularity, returning a shortlist of relevant options rather than raw catalogue data.
“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.
He explained, “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 is continuing testing with a major LLM platform that began in 2025 and is expanding trials with additional platforms, retailers, and brands.
