Evals

Evals provide systematic assessment of AI models, functions, and workflows to ensure quality, reliability, and business alignment.

Overview

The Evals system provides a way to evaluate the performance and quality of your business AI applications. Evals can:

• Assess model performance on specific business tasks
• Validate function outputs against expected business results
• Measure workflow efficiency and effectiveness
• Ensure compliance with business requirements
• Track improvements over time

Components

• Metrics: Measure business performance and quality
• Tests: Test business AI components and systems
• Benchmarks: Compare against industry standards

Getting Started

To get started with the Evals collection, explore the following pages:

• Metrics
• Tests
• Benchmarks