What is relativedb?
relativedb answers questions about the future of your relational data. You declare the shape of your tables and links, wire small retriever callbacks over your existing storage, and write a predictive query:
PREDICT NOT EXISTS(orders.*) OVER (90 DAYS FOLLOWING) FOR EACH customers.customer_id
That's 90-day churn for every customer — no feature engineering, no training pipeline, and no temporal leakage by construction.
How it fits together
- RelQL — a SQL-flavored query language for predictions. Parsed and validated against your declared schema. See the RelQL docs.
- Retrievers — the engine never touches your database. All data access goes through callbacks you implement, GraphQL-style. See Retrievers.
- Temporal context assembly — the engine hops your relational graph to build a per-entity context, and guarantees nothing newer than the anchor time enters it. See Temporal correctness.
- Model backends — contexts are scored by a pluggable backend: a transparent history baseline by default, or RT-J, a relational transformer foundation model that predicts in-context. See Model backends.
Three peer libraries
The engine is implemented natively in Python, Java, and Rust — same concepts, same behavior, idiomatic APIs. A shared C++ inference engine serves the RT-J model to all three.
Where to go next
- Installation and Quickstart — first prediction in minutes.
- RelQL tutorial — learn the language step by step.
- How-to guides — churn, ranking, forecasting, custom retrievers, the native model backend.