Pinterest Focuses On AI Cost-Savings, Expands Visual Search Tools

Pinterest is signaling a strategic shift toward greater use of open source AI models, positioning the approach as a key cost-saving measure while it continues to scale its AI-driven discovery and shopping tools. During the company’s latest earnings call, CEO Bill Ready emphasized that open source models are showing strong performance in Pinterest’s core visual use cases, often at substantially lower cost than proprietary large language models.

Pinterest has increasingly leaned on AI across its platform, from personalized recommendations and multimodal search to advertising relevance and product discovery. The recent launch of Pinterest Assistant — an in-app conversational assistant that helps users navigate shopping and inspiration tasks — is part of the company’s effort to deepen engagement and move closer to what analysts describe as “agentic commerce,” where AI can guide and potentially perform parts of the purchasing process.

However, the discussion comes at a moment of investor concern. Pinterest projected a softer holiday quarter, pointing to macroeconomic pressure tied to tariffs impacting home furnishings, a category where the platform is influential in early-stage product research. The company expects fourth-quarter revenue between $1.31 billion and $1.34 billion, slightly below analyst expectations. Following the forecast, Pinterest’s stock dropped more than 20%.

Ready said the company runs ongoing comparisons between leading proprietary AI models and fine-tuned open source alternatives. “We’re seeing orders of magnitude reduction in cost with comparable performance using fine-tuned open source models versus leading off-the-shelf proprietary models,” he told investors, adding that Pinterest intends to move ahead with open source models “at a fraction of the cost.”