Documentation Center

Everything you need to build, deploy, and manage deterministic AI systems with RailMind. From quick start guides to deep API references.

Quick Start

Deploy your first AI Rail in under 5 minutes with our step-by-step guide.

API Reference

Complete REST API documentation with examples for every endpoint and SDK method.

Tutorials

Hands-on guides for common use cases: sentiment analysis, data extraction, content generation, and more.

Core Concepts

Understanding these foundational concepts will help you make the most of RailMind.

Rails

A Rail is a named, versioned AI pipeline with a defined input/output schema, model binding, and guardrail configuration. It is the fundamental unit of deployment in RailMind.

Schemas

Schemas define the structure of your AI's output. RailMind validates every LLM response against your schema before it reaches downstream systems, ensuring type safety and structural integrity.

Guardrails

Guardrails are safety policies attached to a Rail. They include PII detection, prompt injection filtering, token limits, content moderation, and custom validation rules.

Routes

Routes define how tasks are distributed across different LLM providers. The intelligent router evaluates complexity, cost, and latency to pick the optimal model for each request.

SDK Quick Reference

RailMind provides official SDKs for TypeScript and Python.

TypeScript
import { RailMind } from '@railmind/sdk'; const rail = new RailMind({ schema: './sentiment.json', model: 'claude-3.5', }); const result = await rail.run({ input: "Analyze this feedback", });
Python
from railmind import RailMind rail = RailMind( schema="./sentiment.json", model="claude-3.5", ) result = rail.run( input="Analyze this feedback" )

Changelog

Recent updates and improvements to the RailMind platform.

v2.4.0 March 10, 2026

Added multi-model cascading for intelligent routing. New fallback policies and cost-aware model selection. Dashboard now supports dark/light themes.

v2.3.0 February 18, 2026

Introduced autonomous RAG healing module. Stale vector index detection now runs on configurable schedules. Python SDK reaches feature parity with TypeScript.

v2.2.0 January 25, 2026

OpenTelemetry integration for full distributed tracing. New guardrail plugins for HIPAA and SOC 2 compliance. Performance improvements across the board.