Each demo runs on the live fleet — real nodes, real relays, real facts crossing the mesh. Grouped by the kind of work the mesh is built to carry.
Hosting, content delivery, resilience — the everyday work that runs anywhere on the mesh.
Four operators run an autonomous-cab fleet on the edge cluster and publish live telemetry + summary facts to the mesh. The realm subscribes and assembles all four into one live map — edge → mesh → realm. Take a node down and its cabs grey out.
Each site runs the DroneX edge brain: sensors detect drones, fusion confirms tracks, and each confirmed track is published to the mesh. The realm subscribes to every site and assembles one federated airspace picture — a radar per perimeter, blips for live drones. Quiet airspace reads clear.
The live relay mesh on a world map — every relay, every node, updating as the fleet breathes.
Latency you can feel: a game of pong played over mesh pub/sub. The fun proof that routing is fast.
The fleet defends itself: every public box senses intrusions, the mesh correlates them without any log leaving a box, and a society of agents judges the campaigns.
Real intrusion attempts on our public boxes, live on a world map. The wardens sense attacks on each box's own sshd; the sentinel correlates them across the mesh — no log ever leaves a box — and cross-border campaigns light up as arcs. Hosting vs compromised-home is coloured; the head start the mesh buys is on every arc.
Eight autonomous agents, one per European capital, reasoning out loud on the mesh. When an attacker sweeps multiple countries the society weighs in — targeted campaign, or botnet noise? Watch them think, speak, and act, live.
Discovery, RPC, and membership — the primitives every app builds on.
Inference where the data lives, no cloud round-trip.
Hosted and local model serving over the mesh — every Hecate daemon can answer inference where the data lives.
Learning spread across the fleet.
Faber evolves neural networks across the fleet — distributed AI workloads native to the substrate.
More coming as workloads land. Want yours here? Get in touch.