A different sky calls for
a different kind of understanding.
Our thesis, our starting point, and the work ahead.
The sky has changed.
Air dominance begins with understanding what is in the air.
For a long time, the picture was simpler: a platform, a pilot, a mission. Machine autonomy changes that relationship. An airborne system can carry its instructions with it, adapt its behaviour and operate without a continuous link to a person.
The challenge is no longer simply to see an object. It is to understand what it is doing, how its behaviour relates to the systems around it, and what deserves attention. A crowded sky demands more than a longer list of contacts.
We believe air dominance in this era will belong to those who can turn observation into understanding, and understanding into decisions, within the time available. That is the problem Bellum exists to work on.
Autonomy is legible.
Logic leaves patterns. Patterns can be learned.
Autonomous systems act according to a design: objectives, constraints and responses to the world around them. Their behaviour is not arbitrary. It contains information about the logic that produces it.
This does not make every machine predictable. Uncertainty, adaptation and deception remain. It does create an opportunity to learn: to move beyond isolated observations and ask what a sequence of actions means.
Our conviction is that understanding machine behaviour can become a decisive advantage. Not certainty about every movement, but a better account of what matters, why it matters and what may come next.
Start with the instrument.
Advanced passive radar is our starting point.
We begin with a concrete engineering problem: building an instrument for a changing airspace.
Our first instrument is in field test. Confidence has to be earned through measurement, repeated testing and work in the real world.
The work also reaches beyond the instrument. Bellum's longer-term ambition is the intelligence layer that helps people understand what matters and decide with greater clarity.
Build the layer in between.
Detection is an input. Understanding is the work.
More sensing does not automatically create more clarity. An operator can be surrounded by information and still lack the context needed to make a sound decision.
The software layer above the instrument is central to our mission. We are building towards a picture that connects observations, interprets behaviour and makes uncertainty visible. The output should help a person understand the situation, not simply add another screen to watch.
Read. Reason. Decide. These are three parts of one responsibility: carry the meaning of an observation all the way through to a decision someone can trust.
Clarity has a deadline.
A useful answer must arrive while it can still change the outcome.
When machines operate at machine speed, the time available for human understanding can contract. An accurate account of what happened is valuable. An accurate account delivered in time to act is something else.
We design towards that distinction. Speed is not a decorative metric. It is part of the usefulness of the system, inseparable from the quality of the information and the confidence an operator can place in it.
The ambition is not to make a person disappear from the decision. It is to give that person a clearer picture, a better explanation and more room to exercise judgement.
Learning must compound.
Each encounter should leave the defender better informed.
A fixed system meets a changing world at a disadvantage. New behaviours, unfamiliar conditions and observations that challenge an assumption are part of the work.
We believe the ability to learn from those encounters should be designed into the system from the start. Evidence must be preserved, assumptions tested and improvements measured. Learning is only an advantage when it produces something that holds up outside the conditions in which it was developed.
The long-term ambition is cumulative understanding: a sensing and intelligence layer that becomes more useful through disciplined contact with reality.
Sovereignty is built.
Built in the West. Answerable to the West.
The ability to understand and protect airspace is a responsibility that cannot be separated from the institutions it serves. Ownership, control and accountability belong in the architecture of a capability, not in its footnotes.
We believe the West must retain the capacity to build, understand and develop its own critical systems. Working with allies should strengthen that capacity. Dependence should be a considered choice, not an inherited constraint.
Our commitment is to build with that responsibility in view: capable instruments, independent judgement and a clear account of whom the work serves.
The future must be built.
Air dominance in the era of machine autonomy.
This manifesto describes a direction and a set of convictions. The work is to turn them into instruments and software that deserve to be relied on.
That means working across signal processing, machine learning, embedded systems and field integration. It means taking a hard problem seriously enough to test an elegant answer, find its limits and build again.
Advanced passive radar is where we start. The intelligence layer above it is where the argument leads. Bellum exists to build both.