Tosh Defence
PREDICTIVE MAINTENANCE

SAJAG सजग

Smart Asset Judgement & Availability Guard

Predict the failure. Protect the mission.

<100ms
Prediction Inference Time
0
GPU or Internet Required
6
Vehicle Systems Monitored
100%
On-Premise Sovereign

THE PROBLEM

Indian Army vehicles break down because maintenance follows a calendar, not actual condition. Workshops wait weeks for spares that should have been pre-positioned. Fixed-interval servicing either over-maintains (wasting resources) or under-maintains (missing failures). When equipment fails unexpectedly, spares are often 2,000 km away in a depot - and the vehicle sits idle for weeks. SAJAG changes both: it predicts failures based on how each vehicle is actually used, and tells Ordnance exactly what spares to stock, where, and when.

THE SOLUTION

What SAJAG Does

SAJAG is an AI platform that predicts when military vehicles and equipment will fail, what spares will be needed, and when to stock them - turning reactive breakdown-repair into proactive predict-prevent. It works with data the Army already has: service records, breakdown logs, and deployment histories. No sensors required for Phase 1.

Every prediction is conditioned on terrain and climate. The same vehicle behaves fundamentally differently in Ladakh (-40°C, 5,500m altitude) vs Rajasthan (+52°C desert) vs Northeast monsoon zones. This terrain correlation is SAJAG's core intelligence - no commercial platform models this because no commercial fleet operates across this range. Classical ML algorithms (XGBoost, survival analysis, time-series forecasting) run on standard CPU hardware. No GPU. No LLM dependency. No internet required for any core function.

SAJAG

सजग (SAJAG) - Alert. Vigilant. Watchful. In Indian Army culture, "Sajag raho" (stay alert) is drilled into every soldier from basic training onwards. SAJAG stays alert to equipment degradation so your fleet stays mission-ready.

CAPABILITIES

Key Features

Vehicle Health Scoring

AI-generated 0-100 health score per vehicle and per system (engine, cooling, transmission, electrical, brakes, suspension). Updated after every data ingestion event with historical trend tracking.

Failure Prediction Engine

Survival analysis + classification hybrid predicts failure probability within 7/14/30/60/90 day windows per vehicle per system. Contributing factors and recommended actions included.

Terrain-Weather Correlator

Core differentiator. Adjusts all predictions based on deployment terrain - high altitude, desert, monsoon, plains. Terrain wear factors calibrated from real operational data.

Spares Demand Forecasting

30/60/90-day spares demand forecast per formation. Maps predicted maintenance to spares Bill of Materials. Dead stock identification and periodicity analysis.

Breakdown Pattern Mining

Association rule mining identifies co-occurring failure patterns. "When ALS operates at >4,000m for >90 days, fuel injection system failure follows electrical failure 89% of the time."

Mission Reliability Calculator

Single reliability percentage per vehicle and per formation. What-if scenarios: "If we service these 3 vehicles first, reliability goes from 64% to 86%."

Maintenance Alert Engine

CRITICAL (7 days) / WARNING (7-30 days) / ADVISORY (30-90 days) alerts with recommended actions, spares lists, and acknowledgement workflow.

Data-First Architecture

Works with existing paper records digitised via CSV/Excel upload, web forms, or mobile app. No sensors required. Sensor integration available for future-inducted vehicles.

DEPLOYMENT OPTIONS

Product Tiers

SAJAG OnPrem

Deployed on Army formation's existing server infrastructure. Works on local network with no internet. Data never leaves formation. Suitable for operational deployments and sensitive formations.

SAJAG Cloud

Full stack on NIC Cloud for rapid pilot deployment. All data on Indian government infrastructure. Suitable for initial evaluation and demonstration.

SAJAG Hybrid

Phase 2

Formation-level instances for data collection and local predictions. Command-level aggregator for cross-formation visibility. Ordnance echelon accesses demand forecasts from command level.

WHY NOT A FOREIGN ALTERNATIVE?

Why SAJAG?

Foreign Predictive Maintenance Platforms
SAJAG
Fleet support
Single OEM, controlled environments
Multi-OEM, multi-vintage (1980s to 2020s)
Terrain modelling
None - designed for highways/factories
Built-in terrain-weather correlation across extreme environments
Sensor dependency
Requires IoT sensors from Day 1
Works with manual log data first, sensors optional later
Infrastructure
Cloud-hosted, GPU required
Standard 8-core CPU server, no GPU, no internet
Data sovereignty
Foreign servers, USD pricing
On-premise, sovereign, INR pricing
Supply chain integration
Not included
Spares demand forecasting linked to Ordnance echelon

Every day, military vehicles break down because we maintain them on a calendar instead of their actual condition. SAJAG predicts failures based on how each vehicle is actually used - in which terrain, at what altitude, in what temperature - and tells Ordnance exactly what spares to stock, where, and when.

INTEGRATION

Works With

Existing Army workshop records (CSV/Excel/paper digitisation)
EME and Ordnance echelon supply chain systems
IMD weather data for terrain-climate correlation
OBD-II and CAN bus sensor adapters (Phase 2)
Mobile app for daily exploitation logs

DESIGNED FOR

Who It Serves

Indian Army (EME Corps)
Indian Air Force (Ground Support)
Indian Navy (Vehicle Fleets)
Paramilitary Forces (BSF, CRPF, ITBP)
Defence PSUs (OFB, BEL, BEML)
State Transport Undertakings

Proudly Made in India

Engineered in India with world-class standards. Deployable anywhere in the world. Full source code available for sovereign audit. No vendor lock-in. No foreign dependencies.

Ready to See SAJAG in Action?

Working prototype available within 30 days. Schedule a live demo with our team tailored to your operational requirements. Anywhere in the world.