Tag: RAG
-
Production-Grade AI Agents: RAG, Tool Calling, and Human Approval
A production AI agent needs more than a capable model. It needs permission-aware retrieval, explicit tools, bounded autonomy, and a human approval path for actions with real consequences.
-
AI Application Security: A Practical Guide to Guardrails, Evaluation, and Observability
AI application security is not solved by a system prompt. Protect the application with least-privilege tools, permission-aware retrieval, input and output controls, evaluation, and traces that make failures diagnosable.