Case Study: Notion enhances workspace search with Cohere Rerank

Notion enhances workspace search with Cohere Rerank

Notion, a connected workspace platform, faced the challenge of helping its users quickly find relevant information as their workspaces grew to contain thousands of documents and databases. Their search needed to be more accurate and faster without increasing storage costs from solutions like embeddings, and it required multilingual support for their global user base. To address this, Notion partnered with vendor Cohere to implement its Rerank product.

Cohere's solution integrated the Rerank model directly into Notion's search, which significantly improved the speed and relevance of results. This eliminated the need for costly embedding and vector search in many cases, reducing operational complexity and costs. The implementation led to improved search performance, substantial cost savings, and scalability, contributing to significant revenue growth for Notion as millions of users adopted its AI features.

Notion

Abhishek Modi

Software Engineer