Recommendation SystemConsumer FacingVerified

Reviewed and published by trentmaziarz, March 23, 2026. Discovered and drafted by our automated research pipeline.

Pinterest built a specialized AI model called PinSage that figures out which Pins are related to each other by analyzing the structure of how billions of users have organized content into boards. If many different users have saved both a recipe for sourdough bread and a specific mixing bowl to their cooking boards, PinSage learns those items are related — even if they look nothing alike visually. PinSage was first described in 2018 and its underlying approach still powers Pinterest's related content recommendations, search, and shopping features today.

Details

PinSage is a graph convolutional network — a type of machine learning model designed to find patterns in graph-structured data (where items are connected to each other, like Pins saved to Boards). It was trained on a graph containing 3 billion nodes (2 billion Pins and 1 billion Boards) and 18 billion edges (the save connections between them). The model generates a numerical embedding for each Pin that captures both its visual and contextual meaning. These embeddings serve as foundational inputs for virtually every other machine learning system at Pinterest — including search retrieval, shopping recommendations, ad targeting, and home feed ranking. Successive systems built on this foundation include PinnerSAGE (for user-level preferences), SearchSAGE (for search queries), and OmniSearchSage (for unified multi-modal search).

Products affected

Related PinsPinterest searchPinterest shoppingPinterest AdsPinterest home feed

Sources & Evidence

Company Disclosure

Advancements in Embedding-Based Retrieval at Pinterest Homefeed

Pinterest Engineering Blog·Feb 2025

https://medium.com/pinterest-engineering/advancements-in-embedding-based-retrieval-at-pinterest-homefeed-d7d7971a409e

Company DisclosureVerified

PinSage: A new graph convolutional neural network for web-scale recommender systems

Pinterest Engineering Blog·Aug 2018

https://medium.com/pinterest-engineering/pinsage-a-new-graph-convolutional-neural-network-for-web-scale-recommender-systems-88795a107f48

Academic Paper

Academic PaperVerified

Graph Convolutional Neural Networks for Web-Scale Recommender Systems

arXiv·Jun 2018

https://arxiv.org/abs/1806.01973

Cite this record

Trace Foundation. (2026). Pinterest: Pinterest built a specialized AI model called PinSage that figures out which Pins are related to each other by analyzing the structure of how billions of users have organized content into boards. If many different users have saved both a recipe for sourdough bread and a specific mixing bowl to their cooking boards, PinSage learns those items are related — even if they look nothing alike visually. PinSage was first described in 2018 and its underlying approach still powers Pinterest's related content recommendations, search, and shopping features today (data as of 2026-03-23) [Data set record]. AI Trace. https://www.aitrace.org/r/practice/76a4a86b-82b2-4e91-b2d5-83e3ef5554f7. Accessed October 14, 2026.

Stable link: https://www.aitrace.org/r/practice/76a4a86b-82b2-4e91-b2d5-83e3ef5554f7
Data as of: March 23, 2026
Last verified: March 23, 2026

How to cite AI Trace

Other practices by Pinterest

OtherPinterest maintains an internal program called Responsible AI that covers three areas: building inclusive AI features (the skin tone, body type, and hair pattern tools described in Practice 21), testing its AI systems for unfair bias (using internal tools that measure whether models perform equally well across different groups of users), and developing guidelines for the safe use of generative AI. The program includes red teaming — structured attempts to find flaws or harmful outputs in AI systems before they launch.OtherPinterest is testing AI-powered upgrades to its Boards feature that go beyond simple user-organized collections. In October 2025, Pinterest launched an experiment in the U.S. and Canada where the platform automatically generates outfit ideas from a user's saved fashion Pins ("Styled for you"), creates entirely AI-curated boards based on trending styles ("Boards made for you"), and suggests related products based on what a user has already saved ("Make It Yours").OtherPinterest uses AI to identify diverse skin tones, body types, and hair patterns across billions of images on its platform, then uses those signals to ensure that search results and recommendations reflect a broader range of human appearances. Users can filter fashion search results by body type ranges, skin tone ranges, and hair patterns. The body type technology launched in September 2023, with public user filters rolling out in March 2024; skin tone ranges have been available since 2018.

Have evidence about Pinterest's AI practices? Submit a report.

Report a Sighting →