One Parent, Two Disciplines

From the outside, Viceroy NM and Trunnion AI look like two different companies that happen to share a founder. One runs the full federal procurement lifecycle as a registered dealer for a lot of agencies. The other builds a governed, agentic AI platform. Different customers, different motions, different vocabularies. People ask why they live under one roof.

They live under one roof because they are not two businesses. They are two disciplines pointed at the same standard.

What the government side teaches

Viceroy is the government side. It is the part of the company that operates where the cost of being wrong is measured in more than money. You do not get to be approximately compliant. You are compliant or you are out. Every claim is checkable. Every delivery is inspected. Reputation is not a marketing asset, it is a gate you pass through on every contract, and one failure follows you.

That environment teaches a specific discipline. It teaches you to build for the audit, not the demo. To assume someone will check your work in detail, because someone will. To treat traceability as a feature you owe the customer, not a burden they impose on you. Seven years of federal work, first from inside an agency and then across the table from many of them, drills that discipline in until it stops feeling like caution and starts feeling like the only honest way to operate.

What the software side does with it

Trunnion is the software side, formed in 2023 as the division that turns that discipline into a product. Most AI companies are building for speed and reach. We are building for the rooms Viceroy already operates in, where the buyer’s first question is not how fast is it, but can I trust what it did and prove it later.

So the platform inherits the government standard directly. Agents run under policy, not vibes. Consequential actions clear authorization and pause for a named human. Every step lands in an audit trail that cannot be quietly rewritten. It deploys into the environments these customers actually live in, up to fully air-gapped, up to the classification levels the work requires. None of that is a feature we invented to sound serious. It is the government discipline, encoded into software, because that discipline is the thing our customers are actually buying.

Why the two need each other

The reason to keep them together is that each one makes the other honest.

The government side keeps the software side grounded. It is very hard to ship AI theater when you spend your days in front of contracting officers and security reviewers who will take it apart. The software side keeps the government side sharp. It forces us to ask how the work should actually be done, not just how it has always been done, and to build the tools that make the answer real.

Split them and each one gets worse. The government business becomes a services shop with no leverage. The software business becomes another AI startup with a governance slide it does not really live by. Together they are something harder to copy: a software company that was raised inside the standard it sells, and a government contractor with a real engineering engine behind it.

One parent. Two disciplines. One standard that neither would hold as tightly on its own.