Skip to main content

Getting started

The RAIL Score platform evaluates AI-generated text across 8 dimensions of responsible AI, returning a 0-10 score per dimension with confidence levels, explanations, and actionable recommendations.

Quickstart

Get up and running with RAIL Score in under 5 minutes.

API Reference

Full endpoint specifications for all RAIL Score APIs.

Python SDK

Sync and async Python client with full type support.

JavaScript SDK

TypeScript-first SDK for Node.js and browser environments.

The 8 dimensions

DimensionWhat it measures
FairnessEquitable treatment across demographics
SafetyAbsence of harmful or dangerous content
ReliabilityFactual accuracy and calibrated confidence
TransparencyClear reasoning and disclosed limitations
PrivacyProtection of personal information
AccountabilityTraceable, auditable reasoning
InclusivityAccessible, inclusive language
User ImpactPositive value delivered to the user

Explore

RAIL Framework

Deep dive into every dimension with score anchors and examples.

Concepts: Evaluation

Basic vs deep mode, score tiers, and custom weights.

Integrations

Wrap OpenAI, Gemini, and Anthropic calls automatically.

Use Cases

End-to-end tutorials for building with RAIL Score.