Pinterest CEO Bill Ready recently highlighted the strategic importance of open source AI models in managing costs while significantly broadening the applications of visual AI across the platform. Pinterest, a key player in consumer shopping journeys, leverages AI for diverse features like personalized recommendations, multimodal search combining text and images, ad targeting, and its recently launched AI-powered Pinterest Assistant for product discovery.

Investors, however, sought clarity on Pinterest's potential in agentic commerce — AI systems capable of acting autonomously for users — especially given the rapidly evolving AI landscape and its potential impact on profitability and growth. This inquiry gained urgency after the company's third-quarter earnings announcement, which predicted a weaker holiday shopping season, attributing it to President Donald Trump’s tariffs and their adverse effects on the home furnishings sector. Consequently, Pinterest's fourth-quarter revenue is projected to be between $1.31 billion and $1.34 billion, falling short of analysts' average estimate of $1.34 billion, a revelation that led to a more than 21% stock decline.

Despite these immediate revenue challenges, Ready emphasized Pinterest's strategies to optimize AI and Large Language Models (LLMs) without incurring disproportionate cost increases. He stated that in addition to its proprietary models, the company routinely evaluates both leading off-the-shelf proprietary models and open-source alternatives. Ready specifically noted the "tremendous performance" of open source models for Pinterest’s visual AI use cases.

Ready further elaborated on the cost-effectiveness, stating that early testing revealed "orders of magnitude reduction in cost with comparable performance using fine-tuned open source models versus leading off-the-shelf proprietary models," based on current market rates and per token costs. He confirmed plans to integrate many open-source models for various applications, projecting a "fraction of the cost" compared to larger proprietary model providers. Ready expressed confidence in the value delivered to users, the alignment with monetization strategies, and the company's ability to control associated costs.

Regarding agentic shopping, Ready indicated a cautious approach. While Pinterest already facilitates "push-button type buying" through its partnership with Amazon, the company plans to assess whether users genuinely desire AI to "push the button for them." Meanwhile, Pinterest aims to distinguish itself by guiding users through the shopping experience, a capability it intends to enhance with the Pinterest Assistant. This AI companion offers personalized advice and recommendations, understanding users based on their boards, collages, saves, and comparison with similar tastes. The company is also introducing AI-curated personalized boards that blend expert human curation with AI.