Bill Ready says open source models are powering visual AI with strong performance and lower cost, even as holiday revenue forecasts fall short
Open Source AI: A Strategic Bet for Pinterest
Pinterest CEO Bill Ready emphasized a clear shift in the company’s AI strategy during its latest earnings call: open source AI models are playing a growing role in powering Pinterest’s visual AI capabilities. According to Ready, these models offer “tremendous performance” at significantly lower cost than proprietary alternatives.
In a landscape where AI development often demands massive investments, Pinterest’s strategy stands out. Rather than relying solely on big-name off-the-shelf large language models (LLMs), the company is fine-tuning open source models to match — and sometimes exceed — performance expectations in Pinterest-specific use cases, especially around visual search and discovery.
“We’re seeing orders of magnitude reduction in cost with comparable performance,” said Ready, highlighting cost-efficiency as a critical competitive edge.
Cost Efficiency Amid Soft Revenue Outlook
The embrace of open source AI comes as Pinterest adjusts expectations for the holiday season. The company’s Q4 revenue projection — between $1.31 billion and $1.34 billion — fell below the average Wall Street estimate of $1.34 billion. The cause? A slowdown in the home furnishings category, worsened by tariffs linked to former President Donald Trump’s trade policies.
The result: Pinterest shares plunged over 21% on the news.
Yet Ready remained optimistic, stressing that AI can offset costs and still create value through innovation. Open source tools allow Pinterest to scale new experiences without adding major new expenses to its infrastructure.
Driving Visual AI with Open Source Power
Pinterest’s strength lies in visual discovery, and AI powers many of its features:
- Personalized recommendations tailored to user preferences
- Multimodal search, which blends image and text queries
- Ad targeting that aligns products with user intent
- Pinterest Assistant, the newly launched AI companion
Open source models, fine-tuned in-house, are becoming central to these capabilities — enabling Pinterest to control costs while expanding AI-assisted shopping experiences.
Agentic Commerce: A Measured Approach
With the rise of agentic commerce — where AI acts autonomously to complete transactions for users — investors are asking if Pinterest will enter the space more aggressively.
Ready’s response was measured and user-first. Pinterest already enables one-click purchases via an Amazon integration, but the company is not rushing into full autonomy.
“We’ll see if users want the AI to push the button for them,” Ready noted, signaling a cautious approach based on user demand and trust.
Pinterest Assistant: AI Meets Human Curation
Rather than replacing users’ decisions, Pinterest is focusing on enhancing decision-making. The new Pinterest Assistant interacts with users, providing shopping advice and personalized ideas based on:
- Users’ saved boards, pins, and collages
- Behavioral patterns matched to similar users
- A blend of AI and expert human curation
This fusion of technology and human taste is positioned as a unique differentiator, helping Pinterest stand out in a crowded AI landscape.
AI Innovation with Practical Impact
Despite financial headwinds, Pinterest’s AI roadmap reflects strategic investment in scalable, cost-effective solutions. By leaning into open source AI, the platform not only reduces dependency on costly LLM providers but also retains flexibility to customize its models.
Ready concluded by expressing confidence in Pinterest’s ability to deliver user value, align AI with monetization, and do it all while maintaining control over infrastructure costs.
