Guardrails
Controls that constrain what an AI system can say or do, input/output filtering, policy checks, least-privilege tool access, so it behaves safely in production.
Related terms
Large Language Model (LLM)
Retrieval-Augmented Generation (RAG)
Vector Database
Embedding
Fine-tuning
Prompt Engineering
Need the team behind the term?
Scope your need →