The Defense Intelligence Agency's chief AI officer expects the next few years to bring a shift from individual AI tools to networks of digital agents coordinating across military commands. Maj. Gen. Robert Kinney described the vision Wednesday during a panel at the DODIIS Worldwide conference in Tampa, Florida.
DIA is currently running a 90-day "sprint" to build out an AI enterprise platform service, Kinney said. The agency is also developing a model context protocol, or MCP, to give AI tools "a more universal way" to access intelligence data. MCP is the technical standard that connects AI systems to external data sources.
"Building on top of that, though - and I've estimated, you know, I'm not going to overpromise and underdeliver per se - but thinking about it in the next year, building agents and then agents-with-agents is the direction that we want to go," Kinney said. "And there's a whole other side to that on the compliance side, the security side, the zero-trust side that we're going to work with our CIO teammates on."
For operations-focused professionals, this matters because coordinated AI agents could eventually change how military planning, logistics, and response cycles work.
Agent-to-agent coordination across command structures
Kinney described the operational vision in concrete terms. In a combatant command, a collection management agent could talk to an operations agent, which could then coordinate with a logistics agent or someone on the planning side - all without direct human involvement at each step.
"Imagine a day where … you're in a combatant command and an agent that's a collection management agent is talking to an agent in the 3 [directorate] in operations and fires agents potentially, talking to a contested logistics agent in the 4 [directorate], talking to another agent in the 6 [directorate for command, control, communications and cyber] or the 5 [directorate] in the planning side. That's where this is going," Kinney said.
The vision assumes the underlying infrastructure - the pipes, as Kinney called them - can deliver the needed data to each agent. DIA is currently working on laying that digital foundation. Professionals who manage data flows or system integration will recognize the pattern: agent capability depends entirely on clean, accessible, and well-governed data.
Human oversight in high-stakes missions
Officials are weighing where to place human-in-the-loop and human-on-the-loop controls. Kinney suggested mission areas with reversible outcomes can tolerate a human simply monitoring agents. But for actions with irreversible consequences, a human must stay in the decision path.
"There certainly are, I think, mission areas … that you can accept a bit more risk in and have a human on the loop versus in the loop. But your more irreversible mission areas - you can pick 'em, fires, etc. - you always have a human in the loop as the capabilities get better and better. It's going to have to be there as an agent to check," Kinney said.
This distinction matters for operations staff. If agents handle routine coordination and data retrieval, that changes workload distribution. But if a mission step cannot be undone, the sourcing, the authorizing, and the default response rules will remain human-controlled regardless of how capable the tech gets.
DIA is also thinking about the tradecraft side of agent-to-agent interaction, Kinney said - how to responsibly use the tech without over-relying on it and what parameters agents can operate within.
Why this matters for operations professionals
If DIA's timeline holds, operational workflows within intelligence and defense will gradually shift from manual, sequential coordination between often-siloed functions to partially-embedded agent-driven loops. That shift is structure called MCP courses that update how data integrations are designed, monitored, and audited.
Operations leaders should track three things: how data contracts between agents get standardized, how compliance holds up when agents talk to each other without a human routing every exchange, and how escalation points get defined for irreversible decisions. The people who understand those layers of the system - not just how to toggle an AI tool - will be the ones who can lead where these changes happen.
For daily practice, the practical starting point is understanding MCP the communication standard that lets AI tools pull from external systems and databases. From there, work on defining clear rules for which steps need human approval and which can run automatically. If that architecture gets built well, the agent-to-agent coordination Kinney describes becomes a governance question, not a technology risk. For those working in operations, the useful question is not whether AI agents will arrive, but which of them you will trust enough to act automatically and where you'll keep a human decision check in the loop.
Your membership also unlocks:
