Could AI-generated fashion videos replace human influencers in driving online shopping decisions? Google is betting on it with the latest update to Doppl, its experimental AI try-on app, now enhanced with a fully shoppable discovery feed. This new feed is designed to blend the addictive scroll of short-form video with the precision of personalized e-commerce recommendations that position Google directly against Amazon, TikTok, and Instagram for consumer attention.
The core of the update lies in Doppl’s capability for the generation not just of static outfit previews but dynamic, AI-synthesized videos of real products. Such clips would simulate drape, movement, and texture on a user’s virtual avatar, done through advanced models of fashion visualization. Such systems would often depend on the combinations of garment segmentation, 3D mesh reconstruction, and physics-based cloth simulation to achieve realistic motion. In the case of Doppl, the app takes user-provided style preferences alongside behavioral signals-such as which outfits are tapped, saved, or lingered on-to refine recommendations with time.
Almost every item in the feed links directly to a merchant, collapsing the gap between inspiration and purchase. That of course follows the “shop‑the‑scroll” experience that platforms such as TikTok and Instagram have perfected, but with one key difference: Google’s feed is completely AI‑generated, without human influencers. That choice sidesteps the logistical and cost challenges of creator partnerships while allowing for infinite content generation at scale. But it also follows broader retail tech trends where AI-generated content goes from niche to mainstream in marketing workflows.
The AI video synthesis powering Doppl’s feed likely uses generative adversarial networks or diffusion models fine‑tuned on apparel datasets, enabling the model to render garments under varied lighting, angles, and body poses. By converting static try‑on images into fluid video, Doppl taps into the proven engagement advantage of motion-short‑form video remains the dominant format across social commerce, as 42% of consumers preferring it for influencer‑style product content. Google’s approach essentially automates what creators do manually: to show products in a visually compelling, platform‑native format.
Another lever is personalization. Recommendation engines on e‑commerce sites often mix collaborative filtering with content‑based filtering, but Doppl’s integration of visual AI enables the recommendation of items because of aesthetic similarity, fit predictions, and inferred style clusters. That level of visual‑semantic matching is hard to reproduce without deep computer vision capabilities, which gives Google an underlying technical advantage-if users embrace AI‑only feeds.
The timing is strategic. Social media trends indicate that consumers increasingly begin their product searches on TikTok, Instagram, or YouTube rather than traditional search engines. For Gen Z, over half skip Google altogether for shopping inspiration. In placing a commerce‑ready feed inside Doppl, it means Google is essentially creating its own closed discovery loop-keeping shoppers within its ecosystem from inspiration to checkout. This is particularly critical now that social search continues to erode traditional SEO pathways.
From a competitive point of view, the move echoes TikTok’s introduction of AI‑powered seller tools like Symphony and Smart+, which automate creative production and campaign optimization. Where Google will have the upper hand is in its AI infrastructure and depth of data reach that could make Doppl’s recommendations more accurate and visually rich than competitors. But it also faces the challenge of consumer perception – Sprout Social’s research indicates 46% of users are uncomfortable with AI influencers, which suggests transparency and a human‑touch layer may be part of the requirement for long‑term adoption.
For now, the rollout is restricted to U.S. users aged 18 and above on both iOS and Android, hinting at a cautious but targeted launch. Should Doppl’s shoppable feed gain steam, the model could serve as a blueprint for AI-driven retail experiences across Google’s broader product ecosystem, from Search to YouTube Shopping integrations. For now, it stands as one of the clearest examples yet of how generative AI, personalization algorithms, and short‑form commerce design patterns are converging into a new kind of shopping interface one where model, stylist, and salesperson are all the same AI system.
