01
Interface
Web UI, REST API, CLI, and SSO / OAuth, the entry points through which operators and integrations engage the platform.
- Web UI
- REST API
- CLI
- SSO / OAuth
Overview
What Trunnion AI does
Trunnion AI is the group's enterprise software and applied AI company, formed in 2023 and today a wholly owned subsidiary of Duskbridge, the parent holding company, alongside Viceroy NM. It is and the enterprise-grade, LLM-agnostic multi-agent orchestration platform beneath the entire product portfolio. It coordinates teams of AI agents across large language models, tools, APIs, approval gates, and deployment environments, with governance, classification controls, and human oversight built in from the start.
Trunnion AI is the foundation beneath the full portfolio: the Defense Agent Framework (DAF), Cortrova, Haultro, Zennvue, RealtrAI, FizzSense, Vortiek, and custom builds are all governed Trunnion AI deployments. Because the framework is LLM-agnostic, it can be tuned for any vertical without re-architecting, and it runs on AWS GovCloud, private cloud, on-premise, or in a fully air-gapped environment. For the full platform story, visit trunnion.ai.
- LLM-Agnostic
- 150+ Agent Patterns
- Human-in-the-Loop
- Air-Gap Ready
How it works
A four-layer architecture for governed operations.
02
Orchestrator
The Trunnion AI Engine. Every action flows through a deterministic policy engine, every decision touches an audit log, and every autonomous step has a human checkpoint.
- Task Router
- Context Manager
- Policy Engine
- State Manager
03
Agent Fleet
Domain-specific AI agents that are hot-swappable and sandboxed, configurable across procurement, compliance, research, and contracting workflows.
- Procurement
- Compliance
- Research
- Contract
- Pricing
- Audit
04
Foundation
A governed base layer of audit logging and human checkpoints, engineered to NIST 800-53 and CMMC Level 2 control alignment on FedRAMP-aligned cloud architecture (no authorization held).
- Audit Log
- Human Checkpoints
- NIST 800-53
- CMMC L2-Aligned
- FedRAMP-Aligned
Capabilities
Inside Trunnion AI
- LLM-agnostic by design
- Trunnion AI is not tied to a single model provider. It orchestrates across LLMs, including OpenAI and Anthropic Claude, so deployments can use the right model for the workload and swap providers without re-architecting.
- Human-in-the-loop controls
- Every autonomous step has a human checkpoint. Operators stay in command of consequential actions while agents handle the research, drafting, and monitoring.
- Full audit logging
- Every decision touches an audit log. The complete chain of agent actions is recorded for review, accountability, and compliance evidence.
- Compliance hooks
- Policy-aware execution with NIST 800-53 alignment built into the foundation layer, so governance is enforced by the platform rather than bolted on.
- Flexible deployment
- Designed for deployment on AWS GovCloud, private cloud, on-premise, or fully air-gapped infrastructure. Any classified use requires a program-provided accredited environment, appropriately cleared personnel, and written authorization.
- Configurable agent fleet
- Domain-specific agents are hot-swappable and sandboxed, configurable across procurement, defense, manufacturing, and logistics workflows.
Trunnion AI is built on Trunnion AI, the group's multi-agent orchestration framework. Explore the fullproduct portfolioor learn how we buildsoftware & applied AIfor government and defense.
Frequently Asked
Common questions
What is Trunnion AI?
Trunnion AI is a proprietary multi-agent AI orchestration framework built by Trunnion AI, a wholly owned Duskbridge subsidiary alongside Viceroy NM. It powers the Trunnion AI software portfolio and is configurable for procurement, defense, manufacturing, and logistics domains. Features include human-in-the-loop controls, full audit logging, compliance hooks, and deployment on AWS GovCloud, private cloud, or on-premise environments.
What does LLM-agnostic mean for Trunnion AI?
LLM-agnostic means Trunnion AI is not locked to a single model provider. It orchestrates across large language models, including OpenAI and Anthropic Claude, so a deployment can select the right model for each workload and change providers without re-architecting the system.
What is multi-agent AI orchestration?
Multi-agent AI orchestration is an architecture where multiple AI agents work together autonomously to complete complex tasks. Trunnion AI coordinates specialized agents for procurement research, compliance monitoring, document generation, contract analysis, and operational intelligence, with human-in-the-loop controls and full audit logging at every step.
Can Trunnion AI run in a classified or air-gapped environment?
Trunnion AI is designed for AWS GovCloud, private-cloud, on-premise, and fully air-gapped deployment. Classified use through the Defense Agent Framework additionally requires a program-provided accredited environment, appropriately cleared personnel, an approved security boundary, and written authorization. Viceroy NM does not represent that it holds a facility clearance or cleared-personnel status.
Which Viceroy NM products run on Trunnion AI?
Trunnion AI powers the full portfolio of vertical products and build paths: DAF (Defense Agent Framework), Cortrova (manufacturing ERP), Haultro (waste and fleet operations), Zennvue (creative event ERP and marketplace), RealtrAI (real estate AI), FizzSense (CO2 and IoT logistics), Vortiek (travel intelligence), and custom software builds.