Every insight traceable to source
Enterprise healthcare buyers require more than aggregated metrics — they need to know where data came from, when it was refreshed, and how it flowed into graph products and analytics.
Care-Intel records provenance across the full stack: cataloged datasets, Airflow pipeline runs, Debezium CDC events, Spark transforms, Apache AGE graph builds, and subgraph packaging to MinIO.
Lineage you can audit
- Dataset catalog — source URLs, cadence, last successful ingest, and freshness status in the Monitor API state store.
- Pipeline observability — DAG run history, failure alerts, and operational logs surfaced in Super Admin.
- Graph build jobs — schema execute jobs, bulk edge loads, and render snapshots tied to warehouse tables.
- Subgraph deliveries — SKU, node/edge counts, artifact SHA-256, and export timestamps for every customer slice.
Governed refresh, not stale copies
Post-ingest hooks refresh active subgraphs when upstream tables change. Buyers receive artifacts with documented lineage — not orphaned graph dumps disconnected from the warehouse.
Walk through lineage in a demo
