EDB Postgres AI Benchmarks Reveal Lead Over Vector and Lakehouse Platforms

The benchmark study, commissioned by EDB, evaluated performance across normalized enterprise hardware to determine how different database architectures handle agentic workloads. While common enterprise setups rely on copying data into separate vector stores or lakehouses, these methods often suffer from performance degradation and synchronization delays at scale. EDB’s approach maintains data in a single, sovereign foundation, eliminating the need for ETL processes.

Benchmark Performance and Operational Impact

In core vector search, EDB Postgres AI achieved a 0.911 Recall@10 score, surpassing competitors including MongoDB and Databricks. For agentic retrieval, the system demonstrated a 27-millisecond median latency, maintaining consistency even when combining filtered vector searches with full-text clinical and protocol data. This speed is critical for real-time applications like fraud detection, where response times directly impact transactional outcomes.

Beyond speed, the findings highlight significant price-performance advantages. Because slower platforms consume more compute resources and tokens to execute identical tasks, EDB Postgres AI proved more economical, delivering up to 76x better price performance than Databricks and 34x better than MongoDB. According to William McKnight, president of the consulting group, the platform’s consistency remains its most distinct advantage; while competing systems slowed as scale increased, the Postgres-based architecture maintained steady performance. This structural integration allows enterprises to manage transactions, analytics, and AI workloads within a unified environment, governed at the data layer.