During Tuesday’s earnings call, Pinterest CEO Bill Ready highlighted the potential of open source AI models to help the company keep costs down as it expands its visual AI use cases.
The image pinboard site often serves as the first step in a consumer’s shopping journey and leverages AI technology to power a variety of features, including personalized recommendations, a multimodal search experience that combines text and images, ad targeting, and, most recently, product discovery with the launch of the AI-powered Pinterest Assistant.
But investors wanted to know what Pinterest’s opportunity in agent commerce (AI systems that can act autonomously on behalf of users) might look like as the AI landscape changes rapidly, and how that would impact its revenue and growth potential.
The question has become even more pressing. In announcing its third-quarter results, the company cited President Donald Trump’s tariffs and their negative impact on the home furnishings category, predicting a weaker-than-expected year-end sales season. As a result, Pinterest’s fourth-quarter revenue is expected to be between $1.31 billion and $1.34 billion, compared to analysts’ average estimate of $1.34 billion. The stock price fell more than 21% on Wednesday following the news.
Despite short-term revenue concerns, Pinterest CEO Bill Reddy pointed out how businesses can make the most of AI and LLM (Large Language Models) without significantly increasing costs. In addition to the proprietary model already built into its cost structure, the company regularly tests its leading off-the-shelf models against open source options, Reddy said, and has found the open source model to be promising.
“One of the really, really interesting things we’re seeing is that we’re getting incredible performance from open source models specific to Pinterest’s use case in visual AI,” Reddy told investors. “Given current market rates and costs per token, early tests show that compared to state-of-the-art off-the-shelf proprietary models, using our fine-tuned open source model reduces costs by orders of magnitude with comparable performance.”
He said the company plans to advance a number of open source models for a variety of use cases, and that it can do so at a “fraction of the cost” of larger model providers.
tech crunch event
san francisco
|
October 13-15, 2026
“We feel really good about the value we’re delivering to our users there, our ability to monetize and align that, and our ability to manage those costs and deliver effectively,” Reddy said.
The executive also touched on other areas where Pinterest is looking at how to best leverage AI, such as agent shopping. Reddy said Pinterest already offers “push-button purchases” through a partnership with Amazon, and will see whether users actually want AI to “push their buttons.”
Meanwhile, Reddy suggested that Pinterest’s biggest differentiator is guiding users through the shopping experience. This is something the company hopes to improve with Pinterest Assistant, an AI companion that users can talk to for advice and recommendations. The company says the assistant understands you based on your boards, collages, saves, and comparisons with people with similar tastes.
Pinterest is also rolling out AI-curated personalized boards, which Ready said combines expert human curation with AI.
