MyGov Hub is an AI-powered government assistant designed to unify fragmented online services into one secure, conversational platform. Instead of juggling multiple apps and websites, citizens can renew licenses, pay utility bills, check summons, or even snap a photo of a document, all within a single hub. With support for both text and voice, MyGov Hub makes public services as easy as chatting with an officer.

© 2025 Goodbye World team, for Great AI Hackathon Malaysia 2025 usage.

Malaysia’s digital government ecosystem is highly fragmented, with nearly 300 different government apps and portals handling various functions. This creates three major challenges:

• Confusion & Wasted Time
  Citizens struggle with multiple apps and inconsistent designs, making services inconvenient and difficult to use.

• Poor User Experience
  Apps often freeze, fail at critical steps (e.g., OTP verification), and exclude seniors and less tech-savvy users.

• High Operational Costs
  Maintaining hundreds of separate platforms is expensive and inefficient for agencies.

Real voices and media reports confirm this problem, citizens complaining about app failures on Reddit, and The Sun reporting how duplication of nearly 300 apps has eroded trust.

MyGovHub’s architecture is a multi-layered, serverless ecosystem designed to process eKYC verification, document analysis, and conversational workflows through a unified government chatbot interface.

It leverages AWS cloud services, MongoDB Atlas, and a React Native frontend, ensuring scalability, fault tolerance, and modularity across all major functional domains.

The architecture is organized into four layers:

1. Layer I - Frontend (User Interaction),
2. Layer II - Backend MCP (Middleware & Application Logic),
3. Layer III - Backend Services (Specialized Microservices), and
4. Layer IV - Database (Data Storage).

The React Native application serves as the main entry point for users. It mimics a WhatsApp-style conversational interface, integrating text, image, and voice interactions.

This layer is responsible for capturing user input and displaying responses. It includes:

1. Mobile Application (React Native + Expo):

   • Provides voice and text input for communication with the chatbot.
   • Integrates a local SQLite database to temporarily cache session data and uploaded media.
   • Uses Axios-based REST API calls to communicate securely with the Backend MCP layer via HTTPS endpoints.
   • Handles user registration, eKYC onboarding, and payment confirmations.
   • Implements OTP verification and facial recognition steps during identity registration.

2. eKYC Onboarding Flow:

   • Users capture a selfie and upload their identity documents (e.g., IC, passport).
   • The app sends these images to the eKYC Backend API, uploading to AWS S3 via signed URLs for verification.
   • Triggers document validation and text extraction service through an API call.
   • Integrates with the OTP Verification API to confirm user identity.
   • Integrates with the Face Recognition API to verify selfie against ID photo.
   • Upon successful verification, user data is stored in the MongoDB database.

3. Chatbot Interface:

   • Users interact with the chatbot to request services, upload documents, and receive responses.
   • Supports multimodal input (text, image, voice, and file upload) and output (text, image, links, and files).
   • Captures user input and sends it to the Backend MCP layer via REST API calls.
   • Displays responses received from the Backend MCP layer.
   • Integrates notification service for reminders, transaction alerts, and OTP messages.

The Backend MCP (MyGov Central Processor) orchestrates all API calls, workflow logic, and integration between the frontend and backend microservices.

It is built entirely on Amazon Bedrock, AWS Lambda functions, API Gateway, and MongoDB Atlas triggers for event-driven workflows.

• Core Responsibilities:

  • Manage session state, route API requests, and invoke appropriate Lambda functions.
  • Handle message processing, LLM integration, and response generation for chat sessions.
  • Bridge communication between frontend (React Native app) and backend microservices.
  • Amazon Bedrock is used for LLM inference and response generation.

• Key Components:

  • Amazon Bedrock: Provide comprehensive, secure and flexible access to leading foundation models to build generative AI app and agent. It powers the conversational AI, handling intent recognition, response generation, and context management.
  • AWS Lambda: Execute backend business logic, including API integrations, database operations, and workflow orchestration.
  • MongoDB Atlas Triggers: Execute Lambda functions in response to database events, such as checking status for government services and payment confirmations.

• Conversation Management:

  • The MCP maintains a session state for each user, tracking conversation history and context. This ensures coherent, multi-turn conversations.
  • Each session is uniquely identified via session_id.
  • Conversation context is stored in MongoDB (short-term) and fetched on each request.
  • AI responses are generated via AWS Bedrock LLM API, processed, and sent back to the app.

This layer consists of specialized microservices that handle distinct functionalities, including eKYC verification, document processing, speech-to-text conversion, payment processing, and notification services.

1. eKYC Services:

   • OTP Verification:
     • Dual-channel (SMS/Email) OTP delivery via AWS SNS and SES.
     • OTP stored in MongoDB with 5-minute expiry.
   • Facial Recognition:
     • Compares selfie from S3 with ID photo using AWS Rekognition.
     • Returns similarity score and verification decision.
   • Document Validation:
     • Uses AWS Textract for OCR text extraction.
     • Validates extracted text against government templates (e.g., license number, NRIC format).

2. Chabot Services:

   • Audio Transcription Service:
     • Uses AWS Transcribe to convert user speech to text.
     • Text is passed to the LLM agent for natural language understanding.
   • Document Analysis and Text Extraction:
     • AWS Textract → AWS Lambda parsing → MongoDB Vstore.
     • Extracted text is summarized or verified through the MCP and displayed to user.
   • Receipt and PDF Generator:
     • AWS Lambda compiles templates and stores PDFs in S3.
     • URLs are shared in chat for download or verification.
   • Notification Service:
     • Scheduled AWS EventBridge + Lambda send reminders (due bills, expiry dates).
     • Notifications are stored in MongoDB and pushed to frontend.

3. Payment Services:

   • Bill Creation (Billplz Integration):
     • Lambda creates bills through Billplz API with metadata (amount, user ID, service type).
     • Frontend receives the payment URL and opens in WebView.
   • Payment Webhook Handling:
     • Billplz sends callback → AWS Lambda verifies payment → MongoDB updates transaction status.
     • Trigger receipt generation via PDF service.

All data persistence is managed by MongoDB Atlas, providing a fully managed NoSQL environment with auto-scaling and TTL-based cleanup.

• Store governmental user profiles, conversation history, payment transactions, and other metadata.
• Data Lifecycle Management:
  • Atlas Triggers automate periodic cleanup (e.g., delete expired OTPs).
  • TTL Indexes ensure transient data (chat sessions, OTPs) auto-expire after 7 days.
  • Scheduled Aggregations perform data normalization for reporting dashboards.
• Integration Points:
  • MongoDB Realm Triggers → AWS Lambda (event-driven updates)
  • Backend MCP fetches and updates documents via Atlas Data API
  • Indexed searches accelerate eKYC lookups and document matching.

• Conversational AI
  Understands typos, slang, and informal language, no rigid menus.

• Smart Q&A
  Not just executing tasks, but also answering questions like an officer.

• OCR + GenAI
  Snap a bill or document, and the system knows what action to take.

• Multilingual Speech-to-Text
  Supports English, Chinese, and Malay for inclusivity.

• Smart Session Handling
  Idle chats auto-expire after 15 minutes, preventing frustration.

• eKYC Verification
  Face match with government IC for trusted identity.

• Data Protection
  Chat history auto-deleted after 7 days.

• Single Device Login
  One active session at a time to prevent hijacking.

• Bank-Grade Security
  HMAC SHA-256 signatures secure every payment transaction.

MyGov Hub follows a Hybrid SaaS Model for government agencies and GLCs:

• Subscription Fees
  Predictable recurring revenue.

• Pay-per-Use
  Scalable costs aligned with actual adoption.

• Integration Setup
  One-time onboarding fee per agency.
  Why agencies pay:

  • Cheaper than building new apps.
  • Lower maintenance and upgrade costs.

• Scalability
  Start with high-demand services (JPJ renewals, TNB bills), expand to more agencies, and eventually offer an API marketplace for banks, insurance, and third-party services.

• For Citizens
  Save time, less confusion, better accessibility.
• For Agencies
  Lower costs, easier upgrades, higher efficiency.
• For Nation
  Stronger digital governance, aligned with Malaysia’s digital-first vision and SDGs.

Welcome Page

Chatbot Conversation

Settings Page

Size Bar

MyGov Hub redefines how citizens interact with the government by turning dozens of disjointed apps into a single AI-powered experience.

"One Hub, All Services"

It’s simple, scalable, and human-friendly, designed to:

• 💡 Save citizens time
• 💰 Reduce government expenses
• 🌐 Accelerate digital-first adoption