Threads
Loading user profile...
The user describes a feature implemented using pgvector to find similar users based on behavior in a B2B SaaS context. This involves creating vector embeddings from user actions over the last 30 days, stored in a PostgreSQL table with an HNSW index for fast querying. The feature aims to enhance collaboration by suggesting users with similar interaction patterns. The user seeks feedback on similar use cases and experiences with behavior-as-vector approaches.
The user describes a feature implemented in their B2B SaaS platform that uses pgvector to find similar users based on behavior rather than content. They detail the process of creating vector embeddings for user behavior, the SQL setup, and the challenges faced, such as handling new users with sparse data and adapting to changes in event types. They express interest in learning about similar use cases of pgvector for personalization at scale.