Perceive. Reason.
Decide on board.

Seven capabilities make up the autonomy brain, and nine problems are where it goes to work. Everything runs on the aircraft: no GPS required, no link to the ground required, and verifiable by design.

Hardware enables.
Software defines.

Different aircraft. Shared software foundations. We design focused modules with clear interfaces so individual components can evolve without rebuilding the whole system.

GPS-less navigation

Navigate beyond GPS

We combine perception, state estimation, and sensor fusion to develop navigation for environments where satellite positioning is unavailable.

Verifiability

Built to be verified

Clear interfaces. Explicit behavior. We design modules that can be inspected and tested individually, then assessed as part of the complete system.

Secure communication

Encrypted end to end

We design end-to-end encrypted communication to protect commands and mission data between their intended endpoints.

Embodied physical AI

Ready for the unexpected

Learned behaviour grounded in the aircraft’s own sensing and dynamics. Where fixed rules end, the system adapts to the situation in front of it: unexpected obstacles, changing conditions, degraded sensors. Always within verified bounds.

Semantic scene understanding

Navigate by meaning

The aircraft understands what surrounds it: buildings, roads, vehicles, people, hazards. This environmental context lets it navigate semantically, take instructions in mission terms, and reason about what it sees rather than only where things are.

Multi-agent orchestration

Orchestrated, decentralized

Aircraft coordinate directly with each other. Tasks are negotiated, shared and re-allocated in the field, without a central controller. Missions continue when a link drops or a unit fails.

Long-term memory & domain knowledge

Memory grounded in the domain

Every flight builds on the ones before it: maps, patterns, prior detections and operator knowledge. Information is learned and grounded in the domain of use, so the system improves across deployments, not just within a single mission.

Detect. Track. Warn.
Then act.

An autonomy brain that flies without GPS, understands its surroundings, remembers what it has seen and coordinates with others changes what a small aircraft can do. These are the problems we are building for.

Early warning

Airspace awareness for critical infrastructure

Airports, power plants, ports, rail hubs, data centres, hospitals and public events need to know what is coming, where and when. We fuse ground sensors and airborne scouts into one live picture, detect small, low and slow objects early, track them, and warn people, operators and civilian infrastructure providers with time to act.

  • Semantic scene understanding
  • Multi-agent orchestration
  • Long-term memory
Interception

Guiding interceptors to the threat

Once a threat is confirmed, seconds matter. The brain guides interceptor drones and ground responders to the right point in space and time. GPS-denied terminal guidance, coordination of several interceptors, and a clean hand-off from the tracking picture. The decision stays with people. The flying does not.

  • GPS-less navigation
  • Embodied physical AI
  • Multi-agent orchestration
ISR

Intelligence, Surveillance and Reconnaissance

Persistent eyes over an area of interest without a pilot per aircraft. Aircraft plan their own coverage, understand what they see, report what matters and keep flying when links or satellites are jammed. Encrypted end to end between aircraft and operator.

  • GPS-less navigation
  • Secure communication
  • Semantic scene understanding
Monitoring & inspection

Infrastructure that inspects itself

Power lines, pipelines, bridges, wind turbines, rail and industrial sites. Aircraft fly the same route again and again, close to structures and without GPS, compare what they see with what they remember, and flag change and damage before it becomes an outage.

  • GPS-less navigation
  • Long-term memory
  • Semantic scene understanding
Search & rescue

Finding people where maps end

Collapsed buildings, forests, caves, ships and industrial halls. Autonomous flight through unknown, GPS-denied spaces, semantic detection of people and hazards, and several aircraft sharing the search, so responders arrive with a location instead of a guess.

  • GPS-less navigation
  • Semantic scene understanding
  • Multi-agent orchestration
Disaster response

A live picture when everything has changed

Floods, wildfires, earthquakes and storms make yesterday’s maps useless. Aircraft build the current picture on the fly, assess damage, locate people and infrastructure at risk, and keep the picture updated as the situation evolves.

  • Embodied physical AI
  • Semantic scene understanding
  • Long-term memory
Perimeter & border

Persistent patrol, autonomous re-tasking

Long perimeters, borders and remote sites cannot be watched by people alone. Aircraft patrol, detect and classify activity, re-task themselves when something happens, and hand a confirmed track to the operator instead of hours of video.

  • Multi-agent orchestration
  • Semantic scene understanding
  • Verifiability
Contested logistics

Deliveries where GPS is denied

Medical supplies, spare parts and equipment to places where satellite navigation is jammed or absent. Disaster zones, underground, indoors, or contested areas. Navigation from onboard sensing, verified mission logic, and secure links.

  • GPS-less navigation
  • Verifiability
  • Secure communication
Maritime & offshore

Ports, platforms and wind farms

Ship- and platform-launched aircraft that patrol harbours, inspect offshore installations and detect approaching vessels and drones. In wind, in spray, and without a reliable GPS fix next to steel structures.

  • GPS-less navigation
  • Embodied physical AI
  • Semantic scene understanding

Something else
that flies?

The brain is not tied to one airframe or one mission. If your problem involves aircraft, autonomy and environments where GPS cannot be trusted, we would like to hear about it.