ossie

Apache Ossie is an incubating open source effort to standardize how semantic models are exchanged across analytics, AI, and BI platforms.

The README describes a semantic fragmentation across today's data stack: the same KPI defined differently across tools, teams spending significant effort manually reconciling definitions, and AI agents producing unreliable outputs grounded in inconsistent business logic. Ossie exists to give those a single, consistent source of truth. GitHub's one line calls it an industry wide specification effort to standardize how semantic metadata is exchanged across analytics, AI, and BI platforms.

Ossie provides a single JSON and YAML based specification that any tool can read and write, aiming to be vendor agnostic. The vision is interoperability, efficiency, and collaboration, with data definitions and value staying consistent as they move between AI agents, BI platforms, and the other tools in the stack. Any tool reading the spec is the whole point of a single source of truth, and the spec formats are meant to keep that barrier low.

The repository is organized into a core spec directory with the specification, the machine readable schema, and accompanying documentation, reference converters to other semantic formats such as dbt, GoodData, Polaris, and Salesforce, examples including a complete TPC-DS model, validation tooling, and project documentation.

The README points to a contributing guide for specification changes, a roadmap document for working groups and planned work, GitHub discussions and issues for conversation, and a Slack community for direct contact. The project carries the Apache incubating label, which signals that it is still finding its shape, and its homepage is ossie.apache.org.

• Source 1 · Project repository
• Source 2 · README