Case Study: Julep achieves real-time AI agent performance with Timescale's Postgres-native platform

Julep powers real-time AI agents in under a second with Timescale

Julep AI, an open-source platform for building agentic AI workflows, faced significant data challenges scaling its production system. They needed a database capable of tracking millions of rapid agent state events, enabling fast vector search for memory, and materializing real-time context, all while keeping their stack simple and Postgres-native. To meet these demands, they turned to the vendor Timescale and its Tiger Data platform.

Timescale implemented a solution using its PostgreSQL platform, which integrated TimescaleDB for time-series data and pgvector/pgAI for vector search. This provided Julep with time-partitioned hypertables for high-throughput event ingestion, continuous aggregates for real-time summaries, and in-database vector generation and search. The results included infrastructure simplification, improved developer velocity, and cost efficiency from automatic data tiering and compression, allowing Julep to turn its data layer into a competitive advantage.

Julep