Dalet Labs

Dalet Labs is launching what it calls the first Generative Commerce app, an iOS experience where every object in a generated scene is a real product you can buy. The pitch is aimed at shoppers who browse to figure out what they want rather than searching for something specific, with the app serving as a visual feed that surfaces styles, products, and combinations a user might not have thought to look up on their own. Early access is invite only through the company's website, which positions this launch as the first consumer-facing output from a team that has been working in stealth on generative models for retail discovery. Under the hood, the company is publishing research on a method it calls Collaborative Generation with Non-Parametric Sampling, a system that generates a visual stream in response to users' interests and consumption trends, where collective user behaviour guides what is generated and newly generated images become reference images for later rounds. In practice that means the feed is meant to evolve with the taste of its audience rather than relying on a fixed catalog of promotional imagery, and the scenes themselves are intended to be shoppable down to individual items. The launch is being announced on X by Omer Bar-Tal, a computer vision researcher who studied at the Weizmann Institute of Science and has been a research lead at Runway after serving as a Founding Scientist at Pika . That background, focused on video and image generation research, gives useful context for why Dalet Labs is approaching commerce as a generation problem rather than a recommendation problem, and why its first product is a visual app rather than a model API.