Pinterest has updated its visual search feature to improve functionality for users who search using images.

The updates include new tools that allow for more precise search refinements and increased accessibility throughout the platform.

One update helps users identify their preferences regarding specific Pins.

Initially available in the women’s fashion category in the US, Canada, and the UK, this feature uses Visual Language Models (VLMs) to generate descriptive words for Pins.

When users select a Pin, an animation highlights it. Relevant descriptive terms then appear to assist in refining the search.

🔗 Source: TechCrunch

🧠 Food for thought

1️⃣ Pinterest’s visual search evolution represents a decade-long strategic differentiator

Pinterest’s new visual search features aren’t an overnight development but the culmination of nearly a decade of technological investment dating back to 2013-2014 during the deep learning revolution 1.

What began as a “moonlight project” by engineer Andrew Zhai has evolved into a core platform capability, now processing what had grown to 600 million visual searches monthly by 2018 2.

This long-term commitment to visual search technology has allowed Pinterest to carve out a distinct position against text-focused search engines like Google, with Pinterest users demonstrating fundamentally different search behaviors. For example, 75% of Pinterest users willingly scroll beyond the first page of results, unlike typical Google users 3.

The platform’s visual discovery approach capitalizes on its unique advantage: Pinterest users often engage without specific intent, browsing for inspiration rather than seeking concrete answers. This creates an entirely different search paradigm 3.

These latest improvements reflect Pinterest’s ongoing strategy to leverage its vast image dataset and strengthen its position as a leading visual discovery platform focused specifically on lifestyle, fashion, and shopping experiences.

2️⃣ Pinterest navigates challenging balance between AI enhancement and AI content concerns

Pinterest’s announcement showcases a fundamental tension in today’s digital landscape: deploying AI to improve user experience while simultaneously combating negative effects of AI-generated content.

While the company is enhancing its visual search with AI-powered Visual Language Models to generate descriptive words for images, it’s simultaneously responding to widespread user complaints about low-quality AI-generated images flooding the platform 4.

Recent Pinterest developments