Skip to main content
The ontology turns raw project databases into shared business objects. Where a pipeline produces a table (customers.parquet), the ontology promotes that table to a typed entity (Customer) with named properties, relationships to other entities, and a stable identity that the rest of the platform (apps, dashboards, queries) can rely on.

Concepts

Entities are company-scoped: one ontology per company, shared across all projects.

How it relates to pipelines

A pipeline writes rows into a project database via a data_sink step. That database can stay as raw data, or you can promote it to an entity. Promoting locks in a primary key, exposes the table to apps and dashboards, and unblocks row-level editing. See Promote a database.

What viewers see

Viewers can browse entities, inspect related records (Customer → their Invoices), and run natural-language queries. They cannot create or modify entities, properties, or relationships—those operations are admin/owner only.

Next steps