Get Started

Deploy Your First AI Rail

Go from zero to production-ready AI governance in under five minutes. Follow the steps below to launch your RailMind instance.

Step 1

Install the RailMind CLI

RailMind ships as a lightweight CLI that runs on macOS, Linux, and Windows (WSL). A single command downloads and installs everything you need.

Bash
# Install RailMind (macOS / Linux / WSL) curl -sSL https://railmind.dev/install.sh | sh # Verify installation railmind --version # → railmind v2.4.0
Step 2

Define Your Business Schema

Create a JSON schema that describes your expected AI output format. RailMind uses this schema to enforce type-safe, deterministic outputs from any LLM.

JSON
// business_logic.json { "name": "customer_sentiment", "version": "1.0.0", "output_schema": { "sentiment": "positive | negative | neutral", "confidence": "number (0-1)", "reasoning": "string" }, "guardrails": { "max_tokens": 512, "block_pii": true, "require_reasoning": true } }
Step 3

Align & Deploy

Point RailMind at your schema and choose your model. The CLI validates, instruments, and deploys your AI rail with full observability built in.

Bash
# Align AI to your business logic railmind align --schema ./business_logic.json --model claude-3.5 # Run a test inference railmind test --input "I love this product!" # → { "sentiment": "positive", "confidence": 0.97, "reasoning": "..." } # Deploy to production railmind deploy --env production --guardrails strict # → Rail deployed at https://api.railmind.dev/v1/customer_sentiment

What's Next?

Your first rail is live. Explore the full documentation for advanced features like intelligent routing, real-time observability, and autonomous maintenance.

Read the Docs Explore Components