Pinterest CEO Bill Ready believes open source AI could be the company’s secret weapon for keeping costs low while improving performance across its platform. Speaking during the company’s third-quarter earnings call, Ready said Pinterest has been testing open source large language models (LLMs) and is seeing “tremendous performance” for its visual AI use cases, at a fraction of the cost of proprietary alternatives.
The image discovery platform, often the first stop for shoppers seeking inspiration, already relies heavily on AI. It powers everything from personalized recommendations and multimodal search to ad targeting and product discovery, including the company’s new AI-powered Pinterest Assistant.
Ready explained that open source models have proven particularly efficient when fine-tuned for Pinterest’s needs, offering comparable quality to commercial models while slashing per-token costs.
“We’re just getting tremendous performance from open source models for Pinterest’s visual AI,” Ready told investors. “In early testing, we’re seeing orders of magnitude reduction in cost with comparable results versus leading proprietary models.”
Pinterest plans to expand the use of open source AI across multiple product areas, which Ready said will help the company “control costs and deliver effectively” while maintaining strong user value and monetization alignment.
The discussion comes amid investor concern over Pinterest’s short-term outlook. The company forecasted fourth-quarter revenue between $1.31 billion and $1.34 billion, slightly below analyst expectations, attributing the slowdown to President Donald Trump’s tariffs that have hurt home furnishings sales. Shares fell more than 21% following the announcement.
Despite the softer guidance, Pinterest is doubling down on innovation, exploring how AI can redefine “agentic commerce”, a concept where AI systems autonomously shop or act on behalf of users. Ready noted that Pinterest already supports “push-button” purchases through its Amazon integration but said the company will move carefully before automating transactions. “We’ll see if users actually want the AI to push the button for them,” he said.
Pinterest’s larger AI strategy centers on enhancing the discovery experience rather than replacing it. The company’s Pinterest Assistant acts as a conversational companion that offers personalized advice based on users’ boards, collages, and saved pins, drawing on behavioral similarities with others who share their tastes.
In addition, Pinterest is rolling out AI-personalized boards that merge expert human curation with algorithmic intelligence. Ready said this hybrid approach, combining creativity and computation, is where Pinterest stands apart.
