Akilapa Guardian

AI Technology

Theft Prediction Engine

Our AI model analyses 47 real-time and historical signals to calculate a live theft risk score for every tracked asset — flagging elevated risk up to 4 hours before a typical theft attempt, allowing preemptive action.

87%Prediction accuracy in 2024 (pre-incident identification)

How it works

The Theft Prediction Engine runs a continuous multi-factor risk model on every tracked asset, combining real-time telemetry with historical crime intelligence and environmental data.

Step 1

Signal Collection

The engine ingests 47 signals including asset location, time of day, speed patterns, historical theft frequency for that road segment, current crime alerts from DSS feeds, weather, and day-of-week patterns.

Step 2

Risk Model Scoring

A gradient-boosted machine learning model trained on 3+ years of Nigerian security incident data produces a risk score from 0–100 for each asset, updated every 90 seconds.

Step 3

Threshold Alerting

When a risk score exceeds configured thresholds (default: 65 for warning, 80 for high, 92 for critical), alerts are dispatched to the operator and, at the critical level, to the SOC for human review.

Step 4

Preemptive Action

SOC analysts review critical-level alerts in real time. Options include contacting the driver, dispatching rapid response, or triggering remote vehicle immobilisation if the vehicle is confirmed unoccupied.

Data inputs

  • Real-time GPS telemetry
  • Nigerian Police Force and DSS crime feed integrations
  • Historical theft incident database (3+ years)
  • Road segment risk index
  • Time-of-day and day-of-week patterns
  • Asset velocity and stop-pattern analysis
  • Weather and visibility conditions

Outputs

Risk Score (0–100)

Continuous theft risk score per asset, updated every 90 seconds

Risk Colour Status

Green / Amber / Red visual indicator on dashboard map

Push Alert

WhatsApp, SMS, and in-app notification when threshold is crossed

SOC Auto-Escalation

Automatic SOC queue entry at critical risk level

Example insight

Asset 'TRK-044' risk score has risen to 78 (High). Asset is stationary for 14 minutes on Sagamu-Benin Expressway at 23:47. 3 incidents logged on this corridor segment in the past 30 days. SOC notified.

Model is retrained quarterly on fresh incident data. Regional models are maintained for Lagos, Abuja, South-South, and North — reflecting distinct crime patterns in each zone.

See Theft Prediction Engine on your fleet

Talk to our security architects about vehicles, fleets, churches, schools, and enterprise assets.