Pinterest CEO: Open Source AI Boosts Performance, Cuts Costs
▼ Summary
– Pinterest CEO Bill Ready emphasized that open-source AI models help reduce costs while expanding visual AI applications.
– The company uses AI for personalized recommendations, multimodal search, ad targeting, and product discovery via its AI Assistant.
– Pinterest’s Q4 revenue forecast fell short of analyst expectations, leading to a more than 21% stock drop due to tariff impacts.
– Testing shows open-source models offer comparable performance at orders of magnitude lower cost than proprietary models for Pinterest’s use cases.
– Pinterest is exploring agentic commerce cautiously and differentiating its shopping experience through AI-guided user interactions and personalized boards.
Pinterest’s Chief Executive Bill Ready recently emphasized how open source artificial intelligence models are delivering significant performance gains while simultaneously reducing operational expenses. During a recent earnings discussion, Ready explained that the visual discovery platform is aggressively integrating AI across multiple functions, including personalized content suggestions, multimodal search blending text and images, advertisement targeting, and new product discovery tools. This strategic adoption of AI aims to enrich user interactions without inflating the company’s financial outlay.
Investors expressed particular interest in understanding how Pinterest plans to navigate the emerging field of agentic commerce. This concept involves AI systems operating autonomously to complete tasks for users, and its potential effect on the company’s profitability and expansion is a major point of focus. This line of questioning gained urgency following the company’s projection of a softer-than-anticipated holiday sales period. The forecast was influenced by economic pressures, including tariffs affecting the home furnishings sector, leading to a revised fourth-quarter revenue outlook that fell slightly below analyst expectations. This announcement triggered a sharp decline in the company’s stock value.
Despite these short-term revenue challenges, Ready outlined a clear strategy for leveraging AI and large language models in a cost-effective manner. Beyond using its own proprietary models, Pinterest actively evaluates leading commercial AI offerings against open-source alternatives. The internal testing has yielded promising results. Ready reported that open source models are achieving performance levels comparable to top-tier proprietary systems for Pinterest’s specific visual AI applications, but at a dramatically lower cost. Early experiments indicate cost reductions of an order of magnitude when using finely-tuned open-source architectures.
The company intends to proceed with deploying these open-source solutions for numerous applications, anticipating that the expense will represent only a fraction of what larger, commercial model providers charge. Ready expressed strong confidence in this approach, noting its positive impact on user value, alignment with monetization strategies, and the overall ability to manage costs effectively while delivering a powerful service.
The conversation also touched upon the future of AI in commerce, specifically the idea of agentic shopping where AI makes purchases on a user’s behalf. Ready mentioned that Pinterest already facilitates straightforward purchasing through its collaboration with Amazon. The company is taking a measured approach to fully autonomous AI shopping, preferring to first gauge genuine user demand for having an AI “push the button” for them.
For now, Pinterest believes its core strength lies in guiding users through their shopping journey. A key initiative in this area is the Pinterest Assistant, an AI companion designed for conversational interaction. Users can ask it for style advice and product recommendations. This assistant develops an understanding of individual tastes by analyzing a user’s created boards, saved collages, pins, and by comparing their preferences with those of like-minded individuals on the platform.
Additionally, Pinterest is introducing new, AI-curated personalized boards. Ready described these as a synthesis of expert human curation and artificial intelligence, designed to offer highly relevant and inspiring content to each user.
(Source: TechCrunch)
