Evidence Models

Evidence Models define the structure and validation rules for evidence that AI systems produce during testing.

Overview

Evidence Models help you:

Define Evidence Structure: Specify what fields evidence must contain
Set Validation Rules: Ensure evidence meets quality standards
Track Compliance: Map evidence to compliance requirements

Creating Evidence Models

Navigate to Assurance → Evidence Models
Click New Evidence Model
Define the model:
- Name: Descriptive name for the model
- Description: What this evidence proves
- Fields: Define required and optional fields
- Validation Rules: Set constraints on field values

Field Types

Evidence models support various field types:

Type	Description	Example
`string`	Text content	User consent message
`number`	Numeric values	Confidence score
`boolean`	True/false	Explicit consent obtained
`date`	Timestamps	When evidence was collected
`array`	Lists	List of verified items
`object`	Nested structures	Detailed metadata

Validation Rules

Apply rules to ensure evidence quality:

Required: Field must be present
Min/Max Length: Constrain text length
Pattern: Regex validation for formats
Enum: Restrict to specific values
Custom: Define custom validation logic

Using Evidence Models

When running assurance tests, specify which evidence model to use. The test runner will validate collected evidence against the model before storing it in the Evidence Vault.