The agentic AI glossary

An Agent Control Plane orchestrates, monitors, and governs autonomous AI agents at scale with visibility, security guardrails, IAM, and policy-as-code enforcement.

An agentic policy engine enforces rules on what AI agents can do, evaluating and controlling tool calls and actions on the execution path. Definition and how it works.

An AI Agent Harness is a secure runtime wrapper that connects an agent model to tools, memory, and APIs with sandboxing, state management, and execution limits.

An AI Firewall enforces content and action policy on the agent execution path, before and during each step, to block prompt injection, unsafe tool calls, and data leakage.

An AI Governance Platform oversees, regulates, and documents enterprise AI to align with legal mandates, ethical standards, and corporate risk tolerances. Full definition.

AI Runtime Inspection and Enforcement intercepts and controls model payloads in real time during live agent loops to block leaks, injections, and unsafe actions.

An AI TRiSM Platform governs trust, risk, and security across the AI lifecycle with explainability, model monitoring, AI-specific security, and data privacy controls.

Business Outcome Orchestration coordinates AI agents around measurable business results, not just model calls. Learn how outcome orchestration differs from agent frameworks.

MCP is an open standard that gives AI models a uniform client-server interface to connect to enterprise tools, databases, and data sources without custom integrations.

Multi-Agent Orchestration coordinates communication, task division, and state management among specialized AI agents executing complex enterprise workflows.

Shadow AI agents are AI agents built and run outside IT governance. Learn the risks of shadow agents and how shadow AI discovery and a control plane bring them under control.

A sovereign AI control plane builds, orchestrates, governs, and observes AI agents in one platform while keeping data inside controlled boundaries. Full definition.

TokenOps is the discipline of governing AI agent token costs with budgets, model routing, and stop conditions. Learn how TokenOps controls agentic AI spend.