Product
Track the full history of every dataset
Complete provenance from Blueprint version to generated dataset — so you always know where your training data came from.
Why it matters
Data provenance for AI compliance
As AI regulation increases, teams need to document exactly what data was used to train each model version. LiteSeed provides a complete, immutable audit trail from schema definition to generated dataset.
Immutable audit trail
Every dataset version records the exact Blueprint version, seed and generation timestamp.
Schema evolution history
Track how your Blueprint changed over time with parent-child version lineage.
Compliance-ready exports
Export provenance metadata alongside datasets for regulatory documentation.
Core capabilities
Dataset version registry
Every generated dataset is stored as a versioned artifact with complete provenance metadata.
- →Blueprint ID + version + seed = complete provenance
- →Generation timestamp, row count and file format recorded
- →Quality score and constraint violation rates stored
- →Linked to experiment runs and optimization history
Blueprint lineage graph
Visual representation of Blueprint version history with parent-child relationships.
- →Parent-child version tree for every Blueprint
- →Field-level diff between any two versions
- →Rollback to any previous Blueprint version
- →Export lineage graph for compliance documentation