This post was created in partnership with InMobi
The next shopping journey may begin before consumers even think to search.
During an ADWEEK House Cannes Lions fireside chat co-hosted with InMobi, Ryan Joe, editor-in-chief of ADWEEK, sat down with Naveen Tewari, founder and CEO of InMobi and Glance. They discussed how agentic AI, visual intelligence, and personalization could transform commerce by helping consumers discover products in more intuitive ways.
From search to intelligent shopping
Tewari described Glance as an AI agent designed to handle much of the product research consumers currently do themselves. Rather than comparing brands, skimming reviews, and searching multiple sites for the best price, he said the system learns users’ preferences over time and identifies relevant products.
“We’ve created an agent that does all of this for you at massive scale. It can look through millions of brands, use the world’s intelligence to essentially figure out the right products for you, make sure that those products are best priced, and find you the best place where you can buy them,” Tewari said.
The company recently announced that Glance will be integrated into Samsung televisions, turning the TV into what Tewari described as an “agentic shopping platform.” Because televisions already inspire purchase decisions through the content people watch, he argued they offer a natural environment for AI-powered commerce.
Why visual AI changes the experience
Building that experience requires multiple AI models working together. Tewari explained that Glance combines a commerce foundation model that understands products, an image-generation model that visualizes products in context, and a user model that continuously learns each shopper’s preferences.
He pointed out that shopping is fundamentally imagistic with a high degree of visual nuance, unlike text-based AI assistants. “If you think about the current AI models, they are all about text,” he said.
So, Glance built a model whose output is an image rather than text. “The idea of generative AI was to essentially combine that product with you—on you—and show how it looks,” Tewari explained.
He added that personalization becomes increasingly important because shopping preferences vary significantly from person to person.
“Shopping is not universal; it is very personal. Therefore, you have to have a user model that truly understands the kind of products you like,” Tewari said.
Tewari also discussed the engineering work required to make consumer AI viable at scale, explaining that the company delayed broad rollout until it reduced image-generation costs enough to support population-scale deployment.
