Pinterest Leverages Open Source AI to Reduce Costs and Expand Capabilities
On Tuesday, Pinterest CEO Bill Ready emphasized the potential of open source AI models to help the company manage costs while expanding its visual AI applications.
The image-sharing platform, often a first stop for consumers beginning their shopping journey, uses AI to power multiple features. These include personalized recommendations, a multimodal search that integrates text and images, ad targeting, and, most recently, product discovery through its AI-driven Pinterest Assistant.
Investors, however, were keen to understand Pinterest’s prospects in agentic commerce — AI systems that can autonomously act on behalf of users — given the fast-evolving AI landscape and how it might influence the company’s revenue growth and profitability.
This question gained urgency after Pinterest forecasted a weaker-than-expected holiday shopping season during its third-quarter earnings call. The company cited tariffs introduced by former President Donald Trump and their negative effect on the home furnishings category. As a result, fourth-quarter revenue is projected between $1.31 billion and $1.34 billion, slightly below the average analyst estimate of $1.34 billion. Following the announcement, Pinterest’s stock dropped over 21% on Wednesday.
Despite these short-term revenue concerns, Ready highlighted strategies to maximize AI and large language model (LLM) use without significantly raising costs. Beyond its proprietary models, Pinterest regularly tests leading off-the-shelf AI models against open source options and has found the latter promising.
“One of the really, really interesting things that we’re seeing is that we are just getting tremendous performance from open source models specifically for Pinterest use cases on visual AI,” Ready told investors. “Given current market rates and per-token costs, in early testing, we’re seeing orders-of-magnitude reductions in cost with comparable performance using fine-tuned open source models versus leading proprietary models.”
He added that the company intends to implement many open source models across different use cases, achieving similar results at a fraction of the cost of larger model providers.
“We feel really good about the value that we’re bringing to the user, our ability to align that with monetization, and our ability to control those costs and deliver that effectively,” Ready said.
Agentic Shopping and AI-Enhanced User Experience
Ready also addressed other potential AI applications, including agentic shopping. Pinterest already offers “push-button buying” through its Amazon partnership and plans to observe whether users actually want AI to make purchases autonomously.
In the meantime, Ready emphasized Pinterest’s unique strength in guiding users through the shopping experience. The Pinterest Assistant AI allows users to ask for recommendations and advice, learning from their boards, collages, saves, and comparisons with others who share similar tastes.
Additionally, Pinterest is introducing AI-curated personalized boards that blend expert human curation with AI insights, further enhancing the user experience.
Also Read:
SoftBank and OpenAI launch new joint venture in Japan as AI deals grow increasingly circular
Soket AI Labs Launches DHRITH, India’s Emotion-Aware Speech Recognition System
