TECHNICAL SCOPE DOCUMENT: "Super App - Driver's Virtual Assistant" 1. Introduction & General Objective The project consists of developing a mobile application focused on financial management, safety, and automation for ride-hailing drivers (Uber, Lyft, InDrive, etc.). The app will run in the background as a "co-pilot", reading screen data to calculate ride profitability in real-time, manage operational costs, and ensure driver safety through intelligent audio recording and emergency alerts. 2. Platform & Base Architecture Operating System: Android Exclusive (Minimum supported: Android 8.0 Oreo). Recommended Tech Stack: Kotlin (Native) or Flutter (with a strict requirement for native Method Channels to handle background services and permissions). Data Storage: Local database (Room/SQLite) for financial history and temporary audio storage. Cloud integration (Firebase/AWS) for emergency audio files and subscription management. 3. Detailed Modules (Features) Module 1: Core Engine (Screen Reading & Calculations) The foundation of the app is triaging ride requests the moment they pop up. AccessibilityService: Capture on-screen text (Platform, Ride Fare, Estimated Time, Distance to passenger, and Trip distance). The app must not use VPNs or network interceptors. Calculation Engine: Real-time math to extract: Earnings per Mile/KM ($/Mile or $/KM). Earnings per Minute/Hour. Automatic deduction of the estimated fuel cost for that specific trip. Overlay (Floating Window - SYSTEM_ALERT_WINDOW): Display the results in a visual card (Decision Traffic Light: Green for profit, Yellow for caution, Red for loss) overlaid on top of the original ride-hailing app. Ride State Logic: The system must visually recognize if the driver tapped "Accept" or if the trip was canceled, to accurately feed the financial history. Module 2: Financial Optimization & Daily Management The app must act as the vehicle's financial manager. Cost Setup: Screens to input current fuel price, vehicle consumption (MPG or Km/L), and fixed overhead costs. Rental Mileage Tracking: Input field for daily/monthly mileage limits for rented cars. The app should issue a visual warning if the driver is driving too many empty miles or exceeding their rental allowance. Dashboard & History: A management panel recording all daily rides, clearly separating Gross Revenue from Net Profit. Live Analytics: A minimized status bar during the trip showing the projected hourly earnings at that exact moment. Module 3: Active Security Hub The main differentiator of the app will be driver protection. Smart Continuous Audio: Background audio recording that triggers automatically when a ride is accepted. Auto-cleanup: Standard audio files must be automatically overwritten or deleted every 48 hours to prevent filling up the device's storage. Emergency Protocol (Offline Wake Word): Integration of an ultra-lightweight, offline voice recognition library (e.g., Porcupine/Picovoice). The app actively listens to the cabin audio. If the driver says a configured "Panic Word" (e.g., a safe word or security code), the app must immediately: Start recording and instantly upload the audio to the cloud. Send the live location via SMS/WhatsApp/Push notification to pre-registered emergency contacts (or a monitoring center). Module 4: User Management & Subscriptions (Paywall) Authentication: Login via Google or Email/Password (Firebase Auth). Subscription Plans: Integration with a payment gateway (Stripe, Google Play Billing, etc.) to unlock "Pro" features (such as Voice AI and extended history). Default Lock: Automatic blocking of premium features if the subscription expires or fails. 4. Non-Functional Requirements & Performance (CRITICAL) To ensure the driver's phone does not lag or freeze during a ride, the following optimizations are mandatory: Sleep Mode (Accessibility): The screen-reading service must pause its processing when the user closes or minimizes the ride-hailing apps. Optimized GPS Cycle: Location requests (GPS) must not run constantly. GPS should be activated and used only when the Emergency Protocol is triggered, significantly saving battery life. Lightweight Voice AI: It is strictly forbidden to use continuous network calls for voice recognition (avoid REST APIs for continuous listening). The trigger word detection must happen locally on the hardware (offline). 5. Expected Deliverables Full source code with technical documentation of the chosen architecture. Installable application (APK/AAB) for Alpha and Beta testing. Technical assistance for the Google Play Store submission, especially in structuring the Privacy Policy (Audio Recording Terms) and filling out the justification forms for the Accessibility permission.