Logic

Meet Logic

Logic turns a written specification into an AI agent that can be called through an API or used through a generated interface. A team defines the expected behavior, inputs, and outputs, then adds examples to test the result. Templates cover tasks such as document extraction, record matching, classification, and content moderation.

The platform versions specification changes, shows differences, and supports rollback or pinned versions. Built-in tests help detect regressions, while execution logs record inputs, outputs, errors, and timing. Logic also handles model routing and provider failover. Batch processing and MCP access let teams use the same agent across datasets, applications, and compatible AI tools while maintaining a defined interface.