EzraDB
Product

Search & AI

Full-text and vectors without another system.

WHAT'S INSIDE

  • Full-text search — tsvector / tsquery / ts_rank, GIN-indexed
  • pgvector-compatible vector type — <-> L2, <#> inner, <=> cosine
  • KNN queries: ORDER BY embedding <-> $1 LIMIT k
  • HNSW / IVF approximate-nearest-neighbor index*

*In development — not yet production-ready.

vector KNN
CREATE TABLE items (id int, embedding vector(3));
SELECT id FROM items ORDER BY embedding <-> '[3,3,3]' LIMIT 5;   -- <-> L2, <#> inner, <=> cosine

THE REST OF THE ENGINE

Transactional engineA serious OLTP core — MVCC, WAL, real Postgres semantics.
Analytical engineA vectorized columnar warehouse living in the same database.
HTAP fabricBoth engines over one MVCC timeline. Pick per table, keep one copy.
Time-series & quantA kdb-class time engine that speaks SQL.
Streaming & CDCData in and out, continuously.
Redis-compatible KV & queuesA cache and job-queue engine inside your database.
In-database computeBring the code to the data — JavaScript today, distributed & GPU next.
Resilience & scaleSpanner-class ambitions, engineered in the open of our roadmap.
Security & operationsRun it like Postgres. Watch it like a modern system.
← Everything inside the engineSee the benchmarks →