> ## Documentation Index > Fetch the complete documentation index at: https://docs.trusys.ai/llms.txt > Use this file to discover all available pages before exploring further. # AI Guardrails TRU GUARD enforces proactive guardrails that prevent unsafe, non-compliant, or undesired AI behavior before responses reach users. It acts as a real-time safety and policy enforcement layer for AI applications across prompts, outputs, and agent actions. Descriptive alt text ### Key Capabilities * Real-time input and output guardrails for safety, compliance, and policy enforcement * Configure custom guardrails for PII exposure, prompt injection, jailbreaks, toxicity, and policy violations * Enforce response constraints such as tone, format, length, or allowed content * Apply guardrails across chat, voice, and agentic workflows * Block, modify, or flag responses based on guardrail outcomes * Seamless integration with existing applications via APIs and SDKs ### Use Cases * Prevent unsafe or non-compliant responses in AI copilots and assistants * Protect production systems from prompt injection and jailbreak attacks * Enforce enterprise AI policies consistently across teams and applications * Safeguard agentic workflows and tool executions * Reduce risk exposure while maintaining high-quality AI outputs