The vocabulary of governed AI.
Governed agentic AI has acquired a vocabulary faster than it has acquired agreed definitions. These are the terms Trunnion uses on every page, defined plainly, with the primary standards each one draws on and how the control plane actually enforces it.
Start anywhere.
Each entry gives a definition you can quote, why the concept matters in production, the mechanics of how it is enforced, and the misconception it usually attracts.
Agent governance
The control layer that decides what an AI agent may do, who approves consequential actions, and how every step is proven afterward.
Read the definitionABAC for AI agents
Access control that evaluates attributes of the agent, the resource, and the context on every action, instead of granting a fixed role up front.
Read the definitionHuman-in-the-loop approval gate
A checkpoint where an autonomous agent pauses and a named person approves, edits, or rejects a consequential action before it executes.
Read the definitionHash-chained audit trail
An append-only log where each entry includes the cryptographic hash of the previous one, so any later alteration is detectable.
Read the definitionAI control plane
The shared layer that composes, authorizes, routes, and records agent work, separate from the applications built on top of it.
Read the definitionMulti-agent orchestration
Coordinating several specialized AI agents through a defined workflow, with explicit handoffs, state, and control over the sequence.
Read the definitionLLM-agnostic model routing
Directing each task to an appropriate language model at runtime, so a provider can be changed without rewriting workflows.
Read the definitionAir-gapped AI deployment
Running an AI system inside a network with no physical or logical connection to external networks, including no model API calls.
Read the definitionClassification-aware workflow
An agent workflow that treats data classification as an enforced input to every access and action decision, not as a label.
Read the definitionTool authorization
Checking an agent's permission to invoke a specific capability, separately from its permission to read the underlying data.
Read the definitionReplayable audit record
An audit record complete enough to reconstruct an agent run step by step, including the reasoning behind each action.
Read the definitionEnterprise agentic AI
AI agents deployed to take real action inside an organization's systems, under policy, approval, and audit rather than as an assistant.
Read the definitionFrom definition to deployment.
Every term here describes something the platform enforces rather than something it markets. These pages show the enforcement.
How the control plane works
Public platform concepts and how a reviewed workflow moves from a scoped mission toward a deployment-specific release.
Security and audit
Policy-based access, layered tool authorization, and reviewable records as design patterns; release evidence determines implementation.
Cloud, on-premise, and air-gap
The same governance behavior in all three environments, up to TS/SCI with CAC and PIV identity.
Platform versus framework
Why a framework leaves policy, approval, and audit to you, and what a control plane enforces centrally.
Trunnion versus Palantir AIP
Model strategy, deployment posture, and commercial model, described factually side by side.
Trunnion versus CrewAI
Governance, audit, and deployment compared with an open-source orchestration framework.
Get started
See the vocabulary enforced on your workflow.
Policy on every action, a human on the consequential moves, and a record you can hand to a reviewer. Request a demo in your environment.