More Designs, Same Standards by Remi Denoyer

How Behavioural AI Keeps Pack Research Fit for a Generative World.
AI has reshaped how packaging gets made — productivity is up, creative output has multiplied, and expectations have followed. The bottleneck is no longer generating designs; it’s knowing which ones will perform. That calls for prediction, not opinion.

In this session, Remi Denoyer shows how Behaviorally built Pack.AI to do exactly that. Unlike generalist LLMs that offer impressions without domain grounding, Pack.AI is a multimodal model trained on thousands of real shopper observations from PackFlash® studies. It uses transfer learning from vision networks to read pack semantics — structure, colour, layout — and fuses that with brand and category context to produce predictions anchored in behavioural data and correlated with sales outcomes across billions of data points.

Remi will walk through how the model understands brand equity on shelf, what it takes to stay predictive rather than just generative, and how the tool fits into a consultant’s everyday workflow — helping insight teams screen more design routes, faster, without lowering their standards. The session speaks directly to this year’s Fit for Change theme: it’s a practical look at how behavioural AI lets practitioners get work done without breaking things and stay effective as individuals, even as the tools and expectations around them keep shifting.

Remi Denoyer

Remi Denoyer is the lead Data Scientist at Behaviorally, the global leader in behavioural insights for packaging, where he drives the company’s AI initiative. A graduate of École Polytechnique (Paris) and UC Berkeley, Remi previously served as founding ML engineer at a YC-backed startup, helping scale the product from seed-stage MVP to a $20M Series A. At Behaviorally, he leads the research and development of Pack.AI — a proprietary multimodal model that fuses packaging imagery, on-pack claims and real shopper behaviour to deliver predictive, data-driven guidance for global CPG brands. He works directly with clients such as L’Oréal and Pernod Ricard, and is a regular advocate for advancing AI literacy across the insights profession. He is based in New York.