Skip to main content

The C2 Copilot

Talk to DomeCommand in plain language. The C2 Copilot reads the picture, weighs the options and does the busywork, while every decision that matters stays yours.

The hardest part of a drone fight isn't running the math, DomeCommand's planner does that continuously. It's everything around the decision: reading a chaotic picture, recalling what you have left, comparing two plans honestly, building the right drill to test a worry. The C2 Copilot is an embedded agent that does that work alongside you, on command, in plain English.

How it works​

The Copilot is agentic: it doesn't just answer questions, it takes actions on your behalf, within strict limits. Given a plain-language request, it decides what to look at, does the work, and reports back. Under the hood it can perceive the live engagement, run the planner, author and replay scenarios, and summarise results, the same operations available in the console, now drivable by conversation.

What it cannot do is the one thing that matters: it never commits an action on its own. Composing a plan is not the same as approving it; the Copilot can draft, compare and recommend all day, but a plan only goes live through the approval gate, which traces back to a human.

What you can ask it to do​

Think of the Copilot as five kinds of help:

Read the situation​

  • "Brief me on what's happening right now."
  • "What's the story with that contact in the north-west?"
  • "How much interceptor inventory do I have left?"

It reads the fused picture and your inventory and tells you, in a sentence, what you need to know, without you hunting through panels.

Weigh the options​

  • "Draft a plan for the current situation."
  • "Compare the top two plans, which spends less and lets fewer through?"
  • "What would a more aggressive posture cost me?"

It can run the planner with the framing you ask for and lay the trade-offs side by side, so you're choosing between clearly-explained options.

Build and rehearse​

  • "Build a saturation swarm from the north with a comms blackout halfway through."
  • "Run that scenario and tell me how it went."
  • "Compare this drill against the one we ran yesterday."

It composes runnable scenarios from a description, replays them, and reports the outcome, turning "I wonder what happens if…" into evidence.

Brief and report​

  • "Summarise the last five runs."
  • "Which scenarios haven't we tested yet?"

It pulls together results into a clean summary you can act on or share, rather than a wall of raw numbers.

Control the tempo​

  • "Switch to continuous planning and keep plans fresh."
  • "Hold off, I'll approve plans manually for the next few minutes."

It can start, pause and re-tune the planning loop on your word, adjusting how autonomous the system is without ever exceeding the limits you've set.

Why "agentic" matters here​

A chatbot answers; an agent acts. The Copilot's value is that a spoken request becomes real work, the picture read, the planner run, the scenario built, in the time it takes to ask. That speed is exactly what a fast fight demands. And it's only safe because the agent's reach stops precisely where human authority begins: it can do everything except decide.

That boundary is deliberate, and it's the same promise the whole product is built on, the machine carries the speed and the bookkeeping; the operator keeps command.