#MindForge_Param
Bachat AI is a high-performance, intelligent financial ecosystem built to empower users with seamless, privacy-first, and highly intuitive expense management. Leveraging advanced native processing and state-of-the-art Large Language Models (LLMs), the application serves as a comprehensive companion for personal financial tracking, analysis, and insights.
To provide a secure, localized, and context-aware financial management platform that eliminates the friction of manual data entry while maintaining strict user data privacy.
- Privacy-First OCR: Integrates Google ML Kit's Text Recognition to process financial documents, receipts, and platform bills entirely on-device. Sensitive financial data never leaves the user's phone during the extraction phase, ensuring zero-knowledge privacy.
- Instantaneous Processing: Bypasses network latency by executing optical character recognition locally, allowing for real-time receipt scanning.
- Context-Aware Insights: Powered continuously by the Gemini 2.5 Flash Lite model. Designed specifically for low-latency, multimodal throughput, providing users with rapid, smart categorization and financial advice.
- Multimodal Input: Capable of understanding complex, natural language queries regarding historical spending and budget caps.
- Localized Speech Processing: Utilizes robust Native Device API hooks for both Speech-To-Text (STT) and Text-To-Speech (TTS).
- Dialect Optimization: Fully natively configured for both Indian regional contexts (
hi-IN) and English (en-US), ensuring natural user interactions mimicking human conversational flow.
- Resilient Infrastructure: Employs a dual-database model. An embedded SQLite layer guarantees the application opens instantaneously and functions flawlessly—even in dead zones without cell coverage.
- Cloud Synchronization: Seamless real-time state synchronization with Firebase Infrastructure (Firestore data layer and Authentication module), ensuring data consistency across multiple devices seamlessly.
- Dynamic Analytics: Real-time generation of holistic spending graphs and categorized charts, enabling users to maintain macro-level awareness of their financial health.
Bachat AI is structurally split into an ultra-responsive client application and a highly secure proxy layer.
- Framework: Flutter (Dart) — chosen for its ability to compile natively to ARM machine code, resulting in sustained 60-120fps UI performance via Skia/Impeller rendering engines.
- State Management: Provider — ensuring deterministic state hydration across the application tree and strict separation of business logic from the UI.
- Local Storage: SQLite (
sqflite) alongside Shared Preferences for rapid token and cache resolution.
- BFF (Backend-For-Frontend) Proxy: To secure the downstream generative AI access, the application routes AI inference requests through a dedicated Node.js/Express proxy infrastructure.
- Key Vaulting: By managing API tokens on the server layer, the application entirely mitigates the risk of reverse-engineering man-in-the-middle (MITM) attacks and credential leakage on client devices.
- API Obfuscation: The mobile client contains zero proprietary API keys. All core processing logic is tunneled through enterprise secure channels to the proxy layer.
- Identity Verification: Handled natively via Google OAuth pipelines running through Firebase Secure Authentication.
- Data Sovereignty: By keeping optical processing strictly on the device hardware, personal financial habits cannot be intercepted during raw ingestion.
To compile the application for deployment architectures, follow standard procedural builds.
- Java: JDK 21+ (
Eclipse Adoptiumconfigured in system PATH) - SDK: Flutter SDK (^3.11.1 minimum environment)
- Hydrate dependencies from centralized registry:
flutter pub get
- Invalidate older compilation caches:
flutter clean
- Compile the production-ready Android package:
flutter build apk --release
(Note: A dedicated .env file should be orchestrated on the deployment proxy server in production mapping to the Gemini infrastructure.)
© Bachat AI Systems. All rights reserved.