AI-powered vehicle monitoring, diagnostics, fraud detection, and automated service management for Indian fleets.
AutoMind AI is a Next.js prototype of a fully autonomous multi-agent fleet management system. It simulates a real-time AI "operating system" for a 50-vehicle Indian fleet — continuously scanning vehicle telemetry, diagnosing failures, comparing service costs, detecting fraud, and auto-booking repairs — all without human intervention.
The prototype is entirely frontend-driven with no backend or API calls. All agents are pure JavaScript functions that generate realistic, noise-varied mock data to simulate live sensor feeds and AI processing pipelines.
| Layer | Technology |
|---|---|
| Framework | Next.js 14 (App Router) |
| UI | React 18 |
| Animation | Framer Motion 11 |
| Charts | Recharts 2 (Area, Bar, Pie, Radar, Line) |
| Styling | Tailwind CSS 3 + custom glassmorphism CSS |
| Fonts | Space Grotesk + JetBrains Mono (Google Fonts) |
| Language | JavaScript (ES Modules, no TypeScript) |
The app is a single-page client application (app/page.js) with a sidebar navigation that switches between 9 sections. All state lives in page.js, and a background ticker drives the autonomous pipeline every 2.5 seconds.
app/
page.js ← Root SPA, state management, pipeline ticker, auto-refresh
layout.js ← HTML document shell, fonts
components/ ← 14 UI components (pages + reusable widgets)
data/
mockData.js ← Barrel re-export
agents/
dataAgent.js ← Fleet data generator
diagnosisAgent.js ← Anomaly detection
costAgent.js ← Service cost intelligence
fraudAgent.js ← Fraud / overpricing detection
schedulerAgent.js ← Auto-booking engine
networkAgent.js ← Fleet-wide network analysis
The core feature of the prototype is a 5-step agent pipeline that runs automatically every 2.5 seconds, processing vehicles in order of criticality (critical vehicles first):
📡 Scan → 🧠 Diagnose → 💰 Cost Analysis → 🛡️ Fraud Check → 📅 Auto-Book
The StepPipeline component renders this pipeline live at the top of every page, showing the current vehicle being processed and which step is active. Results are stored in storedData and surfaced across all 9 sections of the UI.
Additionally, generateVehicles() is re-called every 4 seconds to simulate live telemetry updates across the whole fleet.
The main dashboard. Shows:
- KPI row — Total Fleet, Critical, Warning, Normal vehicle counts, and average risk score
- Fleet Health donut chart — breakdown by status
- Engine Temperature area chart — 24-hour trend for the selected vehicle
- Battery Distribution bar chart — fleet-wide distribution
- Live Activity Log — real-time feed of pipeline events as they happen
- Vehicle Fleet grid — all 50 vehicles as interactive cards; click any to inspect it; the currently-processing vehicle has a cyan scanning animation
- Customer Engagement Agent panel — notification previews showing auto-generated service reminders
Per-vehicle diagnostic report for the vehicle currently in the pipeline:
- Colored priority banner with risk level, score (0–100), and estimated days-to-failure
- SVG arc risk gauge — color-coded: green → amber → orange → red
- Engine Temp (24h) sparkline chart
- Battery Health (24h) sparkline chart
- Detected Issues list with severity-coded colors
- AI Recommendations list
Compares 10 Indian service centres for the current vehicle's repair:
- 2-column service centre card grid — each card shows rank/badge, name, specialties, distance, wait time, rating, adjusted cost (₹), and % vs market average
- Cost Comparison horizontal bar chart for all 10 centres
- Rating vs Cost bar chart
- AI Recommendation winner card — the algorithmically best option highlighted in green
Fraud shield analysis comparing the recommended service quote against market baseline:
- Verdict card — ✅ SAFE or 🚨 FRAUD / SUSPICIOUS (with shake animation on fraud)
- Price Analysis chart — market average vs selected quote
- Scan Results panel — exact % difference and a plain-language explanation
- AutoMind Fraud Shield branding footer
Shows the automated booking system:
- Latest auto-booking banner — booking ID, vehicle ID, service centre, updating live
- Automation indicator card — confirms zero human intervention
- Booking timeline — vehicle, centre, date, time, assigned technician, duration, priority, status
- Fleet Auto-Bookings list — all bookings generated so far
- Simulated customer feedback — star ratings and status per booking
Fleet-wide pattern detection visualised as a neural network:
- KPI row — Affected Vehicles, Fleet Size, Pattern Confidence %, Regions Flagged
- AI analysis narrative — plain-English description of the detected cross-fleet pattern (e.g., battery degradation cluster)
- SVG neural network graph — all 50 vehicles as orbital nodes around a central AI hub, with animated data particles flowing along edges and pulse rings on affected nodes
- Impact Distribution donut — Affected vs Healthy
- Predictive Savings area chart — cumulative cost savings from early detection
Aggregate fleet analytics:
- KPI row — Total Fleet, Analyzed, Auto Booked, Fraud Detected, Avg Risk Score
- Fleet Health by Brand — stacked bar chart (Critical / Warning / Normal per brand)
- Vehicle Type Mix — pie chart
- Fraud by Brand — bar chart
- Avg Service Cost by Brand — bar chart
- Avg Risk Score by Brand — bar chart
- Region Criticality — bar chart (top 9 Indian cities)
- Activity Trend — dual area chart tracking cumulative Bookings vs Fraud Detected over time
Root Cause Analysis and Corrective Action engine — treats the fleet as a product quality signal:
- KPI row — Patterns Found, Brands Affected, Vehicles Flagged, CAPAs Raised, Escalated
- Brand-level issue bar chart — BattCritical, BattWarn, TempCritical, TempWarn, HiMileage per brand
- Radar chart — Critical / Warning / High Mileage per brand
- RCA finding cards — one card per detected issue pattern (e.g., coolant leak, BMS firmware gap), each with:
- Root cause description
- India-specific context (monsoon roads, Rajasthan dust, hard water corrosion)
- CAPA table — specific corrective actions with priority and status badges
Unified Entity Behaviour Analytics — treats every AI agent as a monitored security entity:
- KPI row — Total Ops Logged, Agents Active, Anomalies, Fraud Blocked, Breaches (always 0)
- System Threat Level gauge — pulsing circle with % risk and level: MONITORING / GUARDED / ELEVATED / HIGH
- Agent Operation count bar chart — operations per agent
- Security Trend area chart — cumulative Threats vs Operations over time
- Authorized vs Anomalies pie chart
- Live UEBA Event Log table — every agent action classified as AUTHORIZED or ANOMALY
| Component | Purpose |
|---|---|
AnimatedBackground |
Full-screen fixed canvas with slow-drifting indigo/violet/cyan radial gradient blobs; subtle mouse parallax |
Sidebar |
Fixed left nav (9 sections), live "Autonomous Processing" panel at bottom with current vehicle, pipeline step dots, and fleet status counts |
StepPipeline |
Compact 5-step pipeline bar shown on every page; active step has scan-line animation; completed steps show ✓ |
GlassCard |
Reusable Framer Motion glassmorphism card; supports hover lift, glow color prop, shimmer line, corner accent |
KpiCard |
Animated metric tile with large number, label, pulsing dot, colored top accent bar; 6 color themes |
VehicleCard |
Fleet grid tile showing ID, brand, type, region, status badge, engine temp, battery %, mileage, mini sparkline |
AlertPanel |
Per-vehicle AI diagnostic report with risk gauge, sparklines, issues list, recommendations |
CostComparison |
10-centre comparison grid with charts and AI-recommended winner |
FraudDetection |
Fraud verdict card with price comparison chart and scan results |
BookingConfirmation |
Auto-booking timeline and fleet bookings list |
NetworkGraph |
SVG neural network visualisation with animated particles and fleet health charts |
DashboardPage |
Multi-chart analytics overview |
RCAPage |
Manufacturing RCA/CAPA findings with brand and radar charts |
UEBAPage |
Agent security monitoring with threat gauge and event log |
Defines all 50 Indian fleet vehicles across 9 brands and 12 cities. Brands: Tata, Maruti Suzuki, Mahindra, Hyundai, Kia, Honda, Toyota, MG Motor, Renault. Models include Nexon EV, Creta, XUV700, Seltos, City, Fortuner, Hector, Kwid, etc. Each call to generateVehicles() adds randomized telemetry noise, generates 24-point temperature and battery trend arrays, and derives vehicle status from thresholds — simulating live IoT sensor feeds.
Exports: generateVehicles, getMostCritical, getFleetSummary, STATUS_COLORS
Multi-signal anomaly detection. Evaluates three signals per vehicle:
- Battery — critical (<25%), low (<40%), borderline (<60%)
- Engine Temp — critical (≥107°C), high (≥92°C), elevated (≥85°C)
- Mileage — extreme (>120k km), high (>90k km), routine (>60k km)
Accumulates a 0–100 score, assigns a risk level (CRITICAL / HIGH / MEDIUM / LOW), estimates days-to-failure, and returns structured issues and recommendations.
Defines 10 Indian service centres with base costs (₹3,400–₹8,200), ratings, wait times, and specialties. Applies vehicle-type multipliers (EV: 1.45×, SUV: 1.15×, etc.) and brand multipliers to produce realistic adjusted quotes. Scores each option with a weighted algorithm: 55% cost efficiency + 25% rating + 20% specialty match to identify the best recommendation.
Exports: getServiceOptions, getMarketAverage, formatINR, starsString
Compares a selected quote against the vehicle-aware market average:
- FRAUD — >28% above market
- SUSPICIOUS — 14–28% above market
- MILD — 4–14% above market
- SAFE — within range or below market
Returns isFraud, severity, pctDifference, message, detail.
Auto-generates confirmed bookings. Assigns appointment dates by priority (CRITICAL/HIGH: next day, MEDIUM: +2 days, LOW: +3 days), time slots (CRITICAL/HIGH always get 10:00 AM), randomly selects a technician from 8 Indian names, generates a BK-XXXXXX booking ID, and sets service duration based on priority. All bookings are flagged autoBooked: true, status: "CONFIRMED".
Computes fleet-wide aggregate stats: pattern confidence (88–97%), regions flagged (5–9), predictive savings (32–58 vehicles), average cost saved (₹2,800–₹4,500). Also generates circular orbital node positions for the SVG network graph visualization.
| Brand | Vehicles | Example Models |
|---|---|---|
| Tata | 8 | Nexon EV, Harrier, Safari, Punch EV |
| Maruti Suzuki | 9 | Swift, Baleno, Brezza, Ertiga, Grand Vitara |
| Mahindra | 7 | XUV700, Scorpio-N, Thar, XUV300 |
| Hyundai | 7 | Creta, Verna, i20, Alcazar, Tucson |
| Kia | 4 | Seltos, Sonet, Carens, EV6 |
| Honda | 4 | City, Amaze, Jazz, WR-V |
| Toyota | 4 | Fortuner, Innova Crysta, Urban Cruiser Hyryder, Glanza |
| MG Motor | 4 | Hector, Astor, ZS EV, Gloster |
| Renault | 3 | Kwid, Triber, Kiger |
Cities covered: Mumbai, Delhi, Bangalore, Pune, Chennai, Hyderabad, Kolkata, Ahmedabad, Jaipur, Chandigarh, Nagpur, Kochi
- Glassmorphism UI — translucent cards with
backdrop-filter: blur, neon accent borders - Dark navy background — deep
#040810to#0C1A2Egradient with drifting ambient blobs - Neon accent palette — cyan
#00D4FF, red#EF4444, green#10B981, amber#F59E0B, purple#A78BFA - Typography — Space Grotesk for UI text, JetBrains Mono for data/code values
- Motion — Framer Motion fade-in-up on card mount, hover lift/scale, shake animation on fraud detection, live scan-line on active pipeline step
npm install
npm run devOpen http://localhost:3000. The autonomous pipeline starts immediately and processes all 50 vehicles in a continuous loop.
- Multi-agent AI orchestration — 5 specialized agents (Data, Diagnosis, Cost, Fraud, Scheduler) chained in sequence, each consuming the output of the previous
- Real-time fleet telemetry simulation — live-updating vehicle data with noise-varied sensor readings
- Autonomous decision-making — bookings are created, fraud is flagged, and recommendations are generated with zero human input
- Indian automotive market specificity — real brands, models, cities, INR pricing, and region-specific RCA context
- Enterprise-grade UI patterns — glassmorphism design system, animated dashboards, KPI cards, network graphs, radar charts, and UEBA security monitoring