How Pinterest Built an AI Engine Users Actually Trust

Pinterest is best known as a place where people go to find ideas, from home décor to weeknight dinners. Behind that simple experience sits one of the most complex artificial intelligence systems in social media.

Pinterest serves 600 million people every month and processes more than 500 petabytes of data, and it needed a way to recommend the right image out of billions of options in real time. The bigger challenge was doing this without becoming just another feed built to keep people scrolling. Pinterest wanted AI that pointed users toward genuine inspiration, not one designed to maximize time spent on the app. That meant building trust and safety into the technology from the start, not adding it later as a patch.

Pinterest has run on Amazon Web Services (AWS) since 2010, which gave the company room to experiment as generative AI matured, according to an AWS case study. Today Pinterest uses Amazon Elastic Kubernetes Service (EKS), a tool for managing computing workloads, to roll out new AI features quickly across the business. More than 10,000 EC2 G5 computing instances handle live recommendations, while another over 600 specialized instances train the underlying models on 18 terabytes of data every day. The result is a system that delivers more than 10 million AI-powered recommendations every second.

One project is Pinterest Canvas, an original AI model that generates and edits high-resolution images. It gives advertisers polished, on-brand backgrounds for their product photos without the manual design work that normally takes. Pinterest also expanded its visual search tools, built with Amazon SageMaker, so the AI can now recognize more than 2.5 billion objects in photos, the case study said.

A new AI assistant lets users search and plan by voice, acting less like a search bar and more like a creative partner, per the case study.

Results Show Growth in Revenue and Users

The approach is paying off. Pinterest reported 17% revenue growth year over year and an 11% increase in monthly active users. Search fulfillment, a measure of how often people find what they’re looking for, improved by 230 basis points. AI now drives about 70% of how people discover content on the platform, according to the case study.

Pinterest Chief Architect Kartik Paramasivam said the payoff goes beyond the numbers.

“For us, trust isn’t just a feature; it’s the foundation of Pinterest,” Paramasivam said, per the case study. “We build, train and leverage AI to filter out content and behavior that is not appropriate for our platform and instead tune our AI to allow our users to easily find what they are looking to explore by personalizing their search and feed.”

Pinterest has signaled it will keep leaning on generative AI to expand tools like Canvas and its voice assistant, while holding the line on its positive-not-addictive philosophy. That balance is worth watching. Many platforms treat responsible AI and business growth as a trade-off, one gained at the expense of the other. Pinterest’s numbers suggest otherwise. For financial services and digital economy companies weighing their own AI investments, the lesson is that user trust and revenue can grow side by side when the technology is built with intent from day one.

“With AWS as our foundation, Pinterest continues to push the boundaries of what’s possible in AI-powered discovery, proving that technology can be both wildly successful and genuinely beneficial to users’ lives,” Paramasivam said, per the case study.

For all PYMNTS AI coverage, subscribe to the daily AI Newsletter.