Onehouse adds vector search to Lakegres on Iceberg, Hudi and Open tables
Onehouse announced vector search inside Lakegres, storing the index natively within existing and new open table formats on object storage. Users can query embeddings via a Postgres endpoint directly on their data lake tables without copying data into a separate database. The feature builds on Onehouse's recent unstructured data processing, letting agents search embeddings from documents, images and video alongside existing lakehouse records.
GoKawiil's interpretation of the reporting above, not reported fact.
By keeping the index in open formats within the customer's own storage bucket, Onehouse is positioning Lakegres as an alternative to copying lake-scale data into application databases like Postgres or OpenSearch just to enable similarity search. The company contrasts this with rival lakehouse vendors whose vector indexes are stored in closed, engine-specific formats, suggesting a lock-in risk if customers switch engines. This could appeal to teams wanting to run AI agents over large historical datasets without duplicating storage or maintaining sync pipelines.
- Lakegres indexes vectors natively inside open table formats like Iceberg and Hudi, avoiding data duplication.
- It exposes a Postgres endpoint for querying embeddings directly on lakehouse data.
- Onehouse contrasts its open-format approach with competitors who store vector indexes in closed, engine-specific formats.
Source: onehouse.ai, 2026-11-18
Published there as: “Lakegres: Vector search on existing Open/Hudi/Iceberg tables”
Read the original report → The summary and analysis above are GoKawiil's own, written from reporting by the source above. Facts and quotes belong to the original publisher.