Case Study: Notion achieves 80% lower search costs and 10B+ vector scale with turbopuffer

Notion needed a search engine capable of scaling beyond 10B vectors to power their next generation of search & AI features

Notion, a productivity and workspace platform, needed a vector search engine that could scale beyond 10 billion vectors to power its next generation of AI features, requiring massive scale with minimal operational overhead. They turned to the vendor turbopuffer to implement its vector database service to meet this challenge.

By implementing turbopuffer, Notion achieved consistent high-performance reads and write peaks exceeding 100,000 operations per second. The solution led to an 80% reduction in cost, which allowed Notion to remove per-user AI charges, and delivered 99.99% uptime with zero performance drops. The turbopuffer team provided highly responsive support and a product roadmap that aligned perfectly with Notion's ambitious growth plans.

Notion

Mickey Liu

Data Engineering Lead