Artifacts

Artifacts in Evalion are the foundational components used to construct comprehensive AI agent testing frameworks.

Evalion organizes artifacts into four main categories that work together to enable thorough agent testing.

Artifacts Components

Artifacts are grouped into the following categories:

Agents

The AI systems being tested and evaluated, including their configuration, connection methods, and behavioral guidelines. Agents serve as the central entity that responds to simulated interactions during test execution.

Scenarios

Specific testing situations or use cases that define realistic user interactions and business contexts. Scenarios provide structured frameworks for measuring how well agents handle different conversation types and user needs.

Personas

User personality types that simulate realistic human behavior during agent testing. Personas define communication styles, patience levels, and behavioral characteristics that mirror user diversity.

Metrics

Measurable criteria used to evaluate agent performance, including both custom semantic metrics and built-in technical measurements. Metrics provide quantitative and qualitative assessments of conversation quality and technical performance.