Milliseconds to Match: Criteo's AdTech AI & the Future of Commerce w/ Diarmuid Gill & Liva Ralaivola

2026-05-09 · Source: The Cognitive Revolution · Field: Technology & Digital — Artificial Intelligence & Machine Learning, Data Science & Analytics, Software Development & Engineering · Depth: Advanced, extended

Summary

Criteo's Diarmuid Gill and Liva Ralaivola, in an interview on May 9, 2026, discussed the intricacies of modern ad technology, emphasizing millisecond-speed recommendation systems and real-time bidding. They explained how Criteo utilizes anonymous user IDs and browsing history to determine ad relevancy and bid values, processing billions of requests daily with low latency. The conversation highlighted the evolution from hand-crafted features to deep learning models and the use of foundation models to generate product and user embeddings, which are then used as inputs for other models. Criteo's partnership with OpenAI aims to combine ChatGPT's broad knowledge with real-time product inventory. The executives also touched upon the company's European roots, its commitment to privacy, and the role of generative AI in democratizing creative content for advertisers, including a self-service product called Criteo Gold.

Key takeaway

For AI Product Managers developing advertising solutions, prioritize building systems that offer transparent value exchange and user control, as exemplified by Criteo's global approach to privacy and real-time bidding. Focus on modular AI architectures that leverage foundation models and embeddings for rapid iteration and performance, while also exploring generative AI to democratize creative content and enhance product discovery experiences for advertisers and consumers alike.

Key insights

Modern ad tech relies on millisecond-speed AI, deep learning, and foundation models for real-time, privacy-conscious recommendations.

Principles

• Advertising must be trustworthy and provide utility.
• Transparency and user consent are paramount.
• Publishing research attracts and retains AI talent.

Method

Criteo employs a hybrid architecture combining LLMs with real-time commerce data. They use deep learning models and foundation models to compute product and user embeddings, enabling rapid, accurate ad bidding and product discovery.

In practice

• Utilize foundation models for "hot start" model training.
• Prioritize user consent and data transparency.
• Explore generative AI for dynamic ad creative.

Topics

• AdTech AI
• Real-time Bidding
• Recommendation Systems
• Deep Learning Embeddings
• Generative Creative AI

Code references

• criteo/deepr

Best for: AI Product Manager, AI Scientist, Machine Learning Engineer, Director of AI/ML

Related on AIssential

• The Hidden Layer of AI: Why Some Brands Get Recommended And Others Disappear — AI systems prioritize inclusion and distributed authority over traditional SEO, creating a new competitive landscape for brand visibility.
• A new generation of ads for the AI era of Search — AI-driven ads provide personalized product guidance and direct offers within Google Search, enhancing transparency and decision-making.
• Enhancing Ad Relevance: Integrating Real-Time Context into Sequential Recommender Models — Integrating real-time context into sequential recommenders significantly boosts ad relevance and business performance.
• OpenAI’s $100 billion ad bet — OpenAI is aggressively monetizing its AI platforms through query-based advertising, aiming to disrupt traditional search and creative industries.
• Stop tracking hypothetical prompts: how to build a prompt set from real buyer queries — Accurate AI visibility tracking requires prompt sets built from real buyer queries, not speculation.
• What AI builders can learn from fraud models that run in 300 milliseconds — AI models can detect fraud in milliseconds by analyzing individual transaction patterns and user behavior.

Open in AIssential →

Editorial summary, takeaway, and curation by AIssential. Original article published by The Cognitive Revolution.