What Does Operational Control of AI Actually Mean?
At a glance: Operational control of AI means a military organization owns its data, its models, its agents and the auditable record of what those agents did, and can keep running that AI when the connection to a remote data center is unavailable. Seekr's SeekrFlow™ platform provides control and explainability over the models and agents themselves. SEMPRE provides the infrastructure they run on — a Total Infrastructure Device™ combining compute, networking, cybersecurity, communications and application hosting in a single system. Under a strategic partnership announced in September 2026, SeekrFlow can operate inside a SEMPRE Total Infrastructure Device, including in disconnected and locally controlled environments.
For a program office, that changes the shape of the work. The AI sits close to the mission systems and data it needs. It stays available locally when reach-back is not appropriate or not possible, and reconnects with enterprise or cloud environments as conditions allow. There is one environment to accredit, secure and sustain rather than a stack the program is holding together itself.
Rob Spalding | CEO
Eight years ago I argued that the United States was building its 5G networks on equipment from a country that might one day want them turned off. The argument was not popular. It was cheaper to buy the foreign equipment, and harder to prove to the right people the security concerns were real.
We are making the same decision about AI right now.
The AI capabilities being demonstrated across the Department of War today assume a working path to a data center somebody else operates, running on infrastructure somebody else built, executing models the government does not hold. Each of those assumptions is reasonable on a good day. None of them is a plan.
For years the AI race was largely a race for the best model. That is already becoming the wrong contest. Models change, and a better one can arrive six months from now. What becomes harder to replace is everything built around it: the data that gives it context, the agents acting on its behalf, their memory, the rules governing them, and the record of what they did.
Owning or licensing a powerful model does not tell a commander whether an agent's recommendation can be reconstructed later, whether mission data stays under government control, or whether the AI can move with the mission as the operating environment changes.
And there is a fourth question. Where does it actually run? A model that depends on reaching a distant data center is only as available as the path to it. Programs solve this by assembling compute, transport, security and hosting from different vendors and then working through the integration, accreditation and sustainment.
Seekr has been working on the first three questions. SeekrFlow lets customers develop and operate models and agents in environments they control, and its explainability tools can trace an output back through the information that influenced it — for agentic workflows, showing the model calls, tool use and the steps taken along the way.
Our partnership with Seekr answers the fourth. SeekrFlow can run inside a SEMPRE Total Infrastructure Device, where compute, networking, cybersecurity, communications and application hosting already operate as one system. We built those pieces ourselves rather than integrating someone else's, which is why they arrive working together on Day 1 instead of becoming the program's integration project.
For a program office, that changes the shape of the work. The AI sits close to the mission systems and data it needs. It stays available locally when reach-back is not appropriate or not possible, and reconnects with enterprise or cloud environments as conditions allow. There is one environment to accredit, secure and sustain rather than a stack the program is holding together itself.
None of this makes the harder questions go away. Deciding which decisions an agent should be trusted with is a command problem, not a procurement one. But those questions are worth arguing about only if the answer is still running when the network is not.
SEMPRE builds Total Infrastructure Devices — compute, networking, cybersecurity, communications and application hosting in one system, built by us rather than assembled from other people's parts. If you are working through where an AI capability will actually run, we are interested in the conversation: info@sempre.ai.
Want to read more about Seekr and SEMPRE’s strategic partnership? Click here.
