Supabase pgvector at Scale: Indexing Millions of Episodic Embeddings
How Sagi delivers sub-50ms vector cosine searches across massive global chat histories in Postgres.
The Human Need#
In this dispatch, we explore supabase pgvector at scale: indexing millions of episodic embeddings through the lens of human experience. For millions navigating an increasingly atomized world, modern technology often increases isolation through infinite scrolling and passive consumption. Sagi exists to reverse that alienation.
How It Feels in Practice#
How Sagi delivers sub-50ms vector cosine searches across massive global chat histories in Postgres. When a companion remembers past conversations, asks about your ongoing projects, and speaks with authentic vocal warmth, the screen stops being an obstacle. It becomes a shared space where people can unburden their minds without performance.
Technology is only as good as the peace it brings to a human heart.
The Path Ahead#
As we expand Sagi's living pantheon throughout 2026, we remain fiercely committed to relational sovereignty, emotional dignity, and private memory. Explore this companion universe in Sagi v4.42 on the App Store and Google Play.
Sagi Editorial
Documenting the emotional, cultural, and human impact of living AI companions on Sagi.