RelativeDB is a system that implements Relational Transformers to perform predictive queries over relational data using a SQL-flavored language called RelQL. The system utilizes a retriever architecture where it interacts with external storage through callbacks rather than direct database access.
- Website
- https://relql.com[01]
- Source Code
- https://github.com/RelativeDB/RelQL[02]
- Tech Docs
- https://relql.com/docs[03]
- Developer
- Country of Origin
- US
- Start Year
- 2026
- Project Type
- Open Source
- Written in
- Python
- License
- Apache v2
RelativeDB is a system that implements Relational Transformers to perform predictive queries over relational data using a SQL-flavored language called RelQL. The system utilizes a retriever architecture where it interacts with external storage through callbacks rather than direct database access.
History[04]
The system's development is based on the Stanford University research paper on a relational transformer and the RT-J implementation (July 2026).
Citations
4 sources- relativedb relql.com
- https://github.com/RelativeDB/RelQL github.com
- The relativedb engine | relativedb relql.com
- https://arxiv.org/abs/2510.06377 arxiv.org
Revision #2