Gold Operational Facts
Summary
Gold is the lakehouse layer for curated facts, features, rollups, and reusable context that downstream automation and review surfaces can consume.
Problem
Raw and canonical data are not always directly useful to operations. A morning brief, campaign audience, identity-resolution workflow, or readiness check may need curated context assembled from many lower-level entities.
Principle
Build reusable facts in gold instead of recomputing complex analytical context inside jobs, UI routes, or review services.
Platform Shape
The lakehouse docs place gold transforms under pipelines/lakehouse/transform/gold. The gold refresh job orchestrates these products through jobs/lakehouse/gold_daily_refresh/run.py. Examples include morning brief context, customer/prospect identity resolution, prospect flags, and other analytics-ready surfaces described in the lakehouse domain docs.
Gold can support operational work, but it should not become interactive state. When operators need low-latency review or repair, publish or adapt gold facts into an OLTP-backed surface.
Source Evidence
docs/domains/lakehouse.mdpipelines/lakehouse/transform/goldjobs/lakehouse/gold_daily_refresh/run.py
Related Reading
docs/engineering/architecture/explicit-publish-steps.mddocs/engineering/architecture/serving-snapshots.md