01
Build
Models · Files · Per-device loadouts
Edge AI for disconnected operations
Mission intelligence for the team in the fight.
- Local AI, mesh, and ATAK on the team's own hardware
- Sync when the mission calls for it
- Wipe when it ends
Data
LOCAL ONLY
SECURED
Mesh
TEAM ONLY
ENCRYPTED
AI
ON-DEVICE
NOMINAL
Mission
EPHEMERAL
WIPE READY
Flagship Research · 2026
The State of the Edge Operator
The technology of war is changing in months, not procurement cycles. Defense leaders must field AI that survives jamming and link loss now, or watch cheaper, faster adversaries set the terms. EdgeLance is building that layer, and the 2026 report lays out the evidence: mission capability belongs on the hardware operators already carry.
The operator is the platform.
Read the Report
What It Is
One operating layer for edge missions.
Mission Workspace
- 10 role-specific views
- 193 API endpoints
- 89 mission components
Local AI
- 35 open-weight models
- Detection, segmentation, speech, reasoning, embeddings
- On-device
EdgeLance Mesh
- 9 transports from Starlink to USB-C Courier
- Store-forward across all
Fleet Control
- Classification-aware MDM
- STIG compliance
- Signed courier delivery
- Duress wipe
Product Screens
Mission flow, not a slide deck.
Camera Intelligence
Contact continuity · Crop tagging · RTSP feeds
Demo Story
Mission pack to field action.
02
Watch
Map · AI directive · Cameras · Readiness · Mesh
03
Move
ISR through local mesh and store-forward
04
Act
Distress · Check-in · Rally — wrist, phone, or laptop
Security Is the Product
Not a checkbox. The architecture.
12 gates. Fail-closed.
Mission Burn
AES-256-GCM key destruction All AI outputs unrecoverable Proof chain survives
Duress
Silent wipe + decoy unlock Covert mesh alert to command MDM-enforced key destruction
12 Gates
API key, WS token, MQTT auth CORS, RBAC, TLS, signing Fail-closed in production
Classification MDM
TS / SECRET / CUI per device STIG continuous compliance Device passport + trust score
Hardware Trust
Firmware provenance scoring Default credential detection Chinese IoT sinkholing
Audit Trail
HMAC-SHA256 signed ledger Tamper-evident Survives mission burn
Why It Is Different
Built for the team, not the enterprise. Local by default. Shared by choice.
124K lines of production code on hardware you already carry.
9 transports. Auto-degrades from Starlink to silent.
Syncs to Lattice, Palantir, ATAK when policy allows. Disconnected when it doesn't.
Mission Burn. Destroy the key, lose the data. Proof survives.
Integrations
Connects to what operators already use.
ATAK / iTAK
Bidirectional CoT XML
Lattice
Entity bridge, selective sync
Palantir AIP
Ontology export, no cloud dependency
Ditto
P2P mesh sync, CRDT conflict resolution
LeapfrogAI
Airgapped LLM via UDS/Zarf
Meshtastic
LoRa mesh, $30/node, store-forward
MAVLink / UAS
Drone telemetry, waypoints, gimbal
Starlink / D2C
9 transport profiles, auto-degrade
USB-C Courier
Signed delivery, wrong-device reject
Cost Calculator
What would cloud AI actually cost?
- All inference on hardware you already own
- Cloud APIs: $15-25 per node per hour
- At scale: hundreds of thousands per month
1 Node / Hour
$16-26
cloud API costs
EdgeLance
$0/hr
local inference
Battalion / Year
$5M+
cloud at 40 nodes
EdgeLance
$0/query
software license, your hardware