Warhacker is not a conference
Defense Unicorns' Warhacker runs June 16-19, 2026 in San Diego. No speakers, no vendor hall, no badge scanning. Roughly 400 participants from government, industry, academia, and nonprofits show up to build and deploy software against real tactical problems. Day 0 is travel and networking. Days 1-2 are warhacking. Day 3 is outbrief.
EdgeLance is there with hardware on the table and a live fleet running.
The demo: mission pack to mesh to burn
Full EdgeLance mission lifecycle on live hardware. A MacBook Pro serves as the command node running the backend, dashboard, and Gemma 2B for local inference. iPhones run as field nodes with camera feeds, detection overlays, and mesh sync. An Apple Watch relays biometric readiness and alert acknowledgment. IP cameras provide fixed-position video with on-device object detection.
Mission flow: Zarf-delivered mission pack containing model weights, detection policies, and mesh configuration. Command node unpacks, loads the model, begins processing. Detections publish to ATAK via CoT. Field nodes sync events and evidence across the mesh. Operator reviews AI recommendations with the source evidence. Mission ends, cryptographic burn destroys all data across every node. One key deletion, mathematically unrecoverable.
Six triggers drive the workflow: mission start, detection threshold, mesh reconnect, evidence promotion, approval gate, and burn command. Eight agents coordinate across the fleet. The entire sequence runs disconnected from the internet.
Fleet delivers. EdgeLance fights.
Defense Unicorns launched UDS Fleet to solve software delivery at the tactical edge. Fleet gets containers, configurations, and updates onto devices in disconnected and denied environments. That is a real solution, backed by $136M in Series B funding.
Delivery is half the problem. Once software reaches the device, something has to orchestrate it: load the right model, route inference, manage mesh, enforce classification, capture evidence, and destroy everything at mission end. UDS Fleet then what is the question. The demo shows both halves working together.
The kit on the table
One MacBook Pro M4 (command node, 38 TOPS, backend + dashboard + Gemma 2B). Two iPhones (field nodes, camera detection, mesh sync). One Apple Watch (biometric relay, alert surface). Two IP cameras (fixed-position object detection). One LoRa radio (mesh fallback). Hardware cost for the cameras and radios: under $5,000. The phones and laptops are devices the team already carries.
Every node managed by EdgeLance fleet layer: software version, model loadout, classification policy, RF emission state, health telemetry. The dashboard shows fleet posture in real time. Any node can be wiped. Any node that goes dark keeps operating and resyncs on reconnect.
Come find us
Warhacker puts builders in the same room. We want operators, engineers, and program managers to break the demo and ask hard questions. What does your approval workflow need to look like? What classification levels matter? What devices are your teams carrying? What integration points with existing C2 would make this useful?
If you are at Warhacker, find us. We will be the ones with the MacBook, four phones, a watch, and a stack of cameras running AI on a folding table. If you are not at Warhacker, reach out directly. The demo runs the same on a single laptop.