Pinterest updates visual search with more AI-powered features

On the 5th May 2025, Pinterest announced a new suite of visual search features designed to enhance how users discover and shop for fashion inspiration – particularly when words fall short. Focused initially on women’s fashion in the U.S., U.K., and Canada, the new tools aim to create a more personalised and intuitive shopping experience by enabling users to search visually rather than through traditional text queries. 

Key updates include an animated glow around elements within images, helping users identify and explore specific parts of an outfit – such as colour palettes or similar garment types. This feature guides users to more precise matches based on visual interest points. 

Pinterest also introduced a refinement bar that allows users to tailor search results to their preferences, including adjustments by colour, style, or occasion. Additionally, users can now initiate visual searches by long-pressing any Pin on their home feed, making it easier to explore looks from anywhere on the platform.

These tools are powered by Pinterest’s Visual Language Models (VLMs) and multimodal embedding technology, which combine image recognition with descriptive text. This generative AI framework allows users to “describe” their preferences visually, without needing to type.

Ultimately, Pinterest’s new features are designed to turn passive scrolling into a curated, actionable, and deeply personal shopping journey, bridging the gap between inspiration and purchase through cutting-edge AI.

Sources: Social Media Today