apache/ossie

What it solves

Apache Ossie addresses the problem of semantic fragmentation in the data stack. It prevents the same KPI from being defined differently across various tools, reduces the manual effort required to reconcile definitions, and stops AI agents from producing unreliable outputs caused by inconsistent business logic.

How it works

It provides a vendor-agnostic, JSON- and YAML-based specification for semantic models. This allows different tools in the data analytics, AI, and BI ecosystem to read and write a single, consistent source of truth for data definitions. The project includes a core specification, reference converters for translating between Ossie and other formats (like dbt and Salesforce), and validation tooling to ensure models adhere to the schema.

Who it’s for

Data engineers, AI developers, and BI analysts who need to ensure consistent data definitions across multiple platforms and AI agents.

Highlights

• Vendor-agnostic standard: Eliminates tool-specific lock-in by providing a common specification.
• Machine-readable: Uses JSON and YAML for easy integration.
• Interoperability: Includes converters for existing semantic formats like dbt, GoodData, Polaris, and Salesforce.
• Validation tooling: Provides tools to ensure semantic models are valid against the schema.