Schema Inference
Automatically determining the shape and types of a system’s data by introspecting its live responses.
Schema inference is how the engine learns the structure of a system it has been pointed at. It calls sample endpoints, evaluates the response shape, infers field names, types, nullability, and nesting, and samples across multiple pages to build a complete picture.
The inferred entities and operations are then mapped onto existing Unified Data Model Nouns and Verbs with a confidence score; low-confidence mappings are flagged for review. The output is a versioned mapping manifest that becomes the contract for the generated connector.
Read next.
Unified Data Model (UDM) A canonical schema of business entities and operations that lets the platform understand the meaning of the data it moves. /glossary/universal-data-model
Autonomous Connector Creation The process by which the engine discovers, builds, and tests a working connector to a system with no human input. /glossary/autonomous-connector
Self-Healing The ability of an integration to detect when an upstream API changes and repair itself automatically, without human intervention. /glossary/self-healing