TROOTH
ALLIED.
A safety system that works even when the internet doesn't.
SafeSakhi turns nearby phones into a real-time emergency network using Bluetooth mesh and on-device AI.
OUR FIRST PROJECT
Most emergency apps rely on cell towers. That's a huge problem if you're stuck in a dead zone, an underground transit system, or a crowded concert where signals jam.
So, we are building SafeSakhi to run completely offline. It uses your phone's hardware to mathematically verify an emergency (like a struggle or a scream) and routes SOS alerts through a local peer-to-peer Bluetooth mesh. No cell service needed.
Servers crash.
Power grids fail.
we engineered a network that survives.
Safety in numbers.
When seconds dictate survival.
Incident Origin
A physical struggle or acoustic anomaly occurs. The device is offline, with zero cellular or Wi-Fi connectivity.
Edge AI Processing
The neural engine captures kinematic shifts and acoustic spikes, processing them through TensorFlow Lite entirely on the hardware. No data is sent to the cloud.
Threat Verification
The on-device model mathematically confirms the signature of a snatch or scream. A localized SOS payload is generated and encrypted with AES-256.
Mesh Activation
CoreBluetooth injects the payload into the surrounding environment. Every nearby device running SafeSakhi silently receives the packet, acting as a secure relay node, bouncing the signal miles away until it hits an internet-connected gateway.
Engineered for the extremes.
01 URBAN DEADZONES
Concrete canyons, elevator shafts, and underground transit systems where conventional cellular penetration drops instantly to zero.
02 HIGH-DENSITY CROWDS
Protests, massive concerts, and gatherings where traditional 5G/LTE towers suffer from bandwidth strangulation and jam entirely.
03 GRID COLLAPSE
Natural disasters and power grid failures rendering standard telecommunication channels and emergency hotlines obsolete.
Stripping it down to fundamental truths.
We don't rely on arbitrary software limits. SafeSakhi analyzes the raw physics of an attack. Our custom models calculate extreme acoustic frequencies (screams) and erratic kinematic velocity (snatches) natively on the device's neural engine.
The Shadow Network.
A live visualization of decentralized packet routing. Each dot represents an offline device. When a threat triggers, the payload cascades through the mesh until it hits a gateway.
Zero Cloud Dependency.
SafeSakhi does not require an active internet connection. It routes encrypted SOS payloads through a localized peer-to-peer Bluetooth mesh network.
Snatch & Scream AI.
Trained in Python and deployed at the edge. The moment a threat is mathematically verified by the device sensors, the mesh activates.
The Mesh Protocol.
Every device running SafeSakhi becomes a secure relay node, building a shadow network that works when the grid collapses.
The Engineering Stack.
Node Initialization
Live output from the CoreBluetooth / BLE execution engine.
First Principles.
If the internet can fail, our solutions cannot depend on it. We build absolute systems for an unpredictable world.
Hardware Optimization
Engineered to run seamlessly in the background with near-zero battery drain using low-energy peripheral states.
Neural Engine
Absolute Privacy
No audio is ever recorded or uploaded. Processing occurs strictly in hardware.