Pinterest has developed a new AI infrastructure in partnership with NVIDIA to enhance its multimodal AI product capabilities, including visual search, content understanding, and safety systems.
The application reports processing over 80bn searches monthly, turning these into valuable AI signals that power discovery and shopping functions.
According to the business, the new system allows Pinterest Assistant to handle 25 times more visual context per request compared to previous capabilities.
Benchmark tests demonstrated roughly 85 times faster response startup and 7.3 times lower overall latency by using precomputed visual representations rather than processing raw images repeatedly, according to the business.
This foundational AI layer supports a wide range of use cases such as ranking algorithms, safety mechanisms, optical character recognition (OCR), and signal generation.
The system aims to combine open-source AI models with Pinterest’s in-house technology and NVIDIA’s comprehensive AI software stack, facilitating faster and more efficient multimodal services according to the company.
Kartik Paramasivam, chief architect at Pinterest said: “Building the next generation of AI-powered discovery means investing in infrastructure that can keep up with the scale and complexity of Pinterest.
“Our collaboration with NVIDIA helps us deliver faster, smarter and more personalized experiences for the hundreds of millions of people who use Pinterest.”
The company emphasizes the ongoing importance of this infrastructure in delivering innovative AI experiences.
The online application is one of the many businesses which is leveraging AI to enhance their overall shopping experience.
Click here to sign up to Retail Gazette‘s free daily email newsletter
