🧠 SuperConfig

Launch your own AI Business Agent in under 3 minutes.
No coding. Just clicks. Fast. Smart. Yours.

🚀 Overview

SuperConfig lets anyone—businesses or individuals—set up a personalized AI Business Agent in minutes. Handle customer support, automate workflows, and manage conversations across multiple channels effortlessly.

💡 Features

⚙️ Easy 3-minute setup
🤖 Personalized AI assistant
💬 Multi-channel customer support
🔐 Secure and customizable

Team: Three Musketeers -Ewen Cheung, Ng Shi Yang, Saw Yong Xuen

Competition: AI Hackathon 2025

🎯 Problem Statement

Most chatbot only handles preset answers, so if a customer phrases a question differently, the bot gets stuck

Staff then have to manually check documents, reply, and update calendars — leading to delayed responses, missed bookings, and frustrated customers

Handle multiple channels (Telegram, future: WhatsApp, Web Chat)
Understand context and maintain conversation memory
Access and search through company knowledge bases
Manage appointments and calendar scheduling
Create and track support tickets
Provide daily business insights and summaries

Why This Matters:

67% of customers expect 24/7 support availability
Poor customer service costs businesses $75 billion annually in the US alone
SMEs lose 30% of potential customers due to delayed response times
Manual ticket management reduces productivity by 40%

🎯 Target Customers

SuperConfig is designed to serve two main groups:

1. Businesses (Our Direct Customers)

We provide SuperConfig as a powerful, easy-to-deploy AI assistant platform for businesses of all sizes—especially SMEs, startups, and service providers who want to automate customer support, streamline operations, and deliver 24/7 personalized service without technical hassle. Businesses can quickly set up their own branded AI agent, integrate it with their existing channels (like Telegram), and manage everything from a user-friendly dashboard.

2. End Users (Business Customers)

These are the customers and clients who interact with the business through SuperConfig—whether they’re booking appointments, asking questions, or seeking support. SuperConfig ensures every end user receives fast, intelligent, and personalized responses, creating a seamless and satisfying customer experience for the business’s audience.

💡 Solution Overview

SuperConfig is an innovative, multi-agent AI customer support platform designed to empower every business with their own personalized AI assistant—set up in just a few clicks. Built for simplicity and power, SuperConfig enables anyone, regardless of technical background, to launch a fully customized support agent that understands your company’s knowledge, manages appointments, handles tickets, and communicates naturally with customers across multiple channels like Telegram. With seamless integration, persistent memory, and advanced language capabilities, SuperConfig transforms customer service into a 24/7, intelligent, and brand-aligned experience—making world-class support accessible to everyone.

Key Capabilities:

🤖 Multi-Agent Intelligence - Specialized agents for scheduling, ticketing, web search, and business insights
📚 Knowledge Base Integration - RAG-powered document search and AI-driven responses
🧠 Conversation Memory - Persistent context awareness across sessions using Mem0
🌐 Real-time Information - Web search capabilities for current events and data
�️ Multi-Language Support - Native conversations in 15+ languages powered by Claude 3 Haiku �📅 Calendar Management - Google Calendar integration for appointment scheduling
🎫 Smart Ticketing - Automated ticket creation, tracking, and escalation
📊 Business Dashboard - Real-time analytics and daily digest reports
🔗 Multi-Channel Support - Telegram (implemented) with extensible architecture
⚡ Asynchronous Processing - Background worker for optimal performance

🛠️ Methodology

Agent-Based Architecture

The system employs a hierarchical multi-agent approach where each agent specializes in specific domains:

Orchestrator Agent: Central coordinator with memory awareness that analyzes user intent and delegates to specialist agents
Scheduler Agent: Manages calendar operations and appointment booking with Google Calendar integration
Ticketing Agent: Creates, updates, and tracks support tickets with priority management
Web Search Agent: Retrieves real-time information from the internet when knowledge base is insufficient
Daily Digest Agent: Generates business insights and summaries from tickets and calendar events

AI Technologies Used:

Large Language Models: Amazon Bedrock (Claude 3 Haiku) for natural language processing
Vector Embeddings: Amazon Titan Embed for semantic search
RAG (Retrieval-Augmented Generation): ChromaDB for knowledge base search
Document Processing: Docling for PDF parsing and chunking
Agent Framework: Strands AI for agent orchestration
Memory System: Mem0 for conversation memory and user context persistence
Real-time Search: Tavily API for current information retrieval

Integration Strategy:

Telegram Bot API for real-time messaging
Google Calendar API for scheduling management
SQLite Database for data persistence and conversation memory
FastAPI for scalable backend architecture
React Frontend for business dashboard
Background Worker for asynchronous message processing
Memory System (Mem0) for conversation context and user preferences

How SuperConfig Works

A customer sends a message on Telegram.
The message is instantly received by our Telegram bot.
Telegram forwards the message to our backend system using a secure webhook.
The backend adds the message to a queue, so nothing gets missed—even if lots of people are messaging at once.
A background worker picks up the message from the queue and gets to work.
The Orchestrator Agent (the “brain” of the system) reads the message, checks past conversations, and figures out what the customer needs.
If needed, the Orchestrator asks specialist agents for help—like the Scheduler for appointments, Ticketing for support issues, or Web Search for real-time info.
The agents can look up information in your company’s knowledge base, check the calendar, or even search the web to find the best answer.
Once the right response is ready, the system sends it back to the customer on Telegram—fast and friendly.
Meanwhile, all conversations and actions are saved, so the agent remembers each customer and can provide personalized service next time.
Business owners and staff can monitor everything, upload new documents, and adjust settings easily from the dashboard.

🏗️ Technical Architecture

System Architecture Diagram

graph TB
    %% === User Layer ===
    Customer[Customer via Telegram]
    
    %% === Frontend Layer ===
    subgraph "Frontend Dashboard (React)"
        Dashboard[Business Dashboard]
        Config[Configuration Panel]
        KB[Knowledge Base Upload]
        Channel[Channel Management]
    end
    
    %% === Communication Channel ===
    Telegram[Telegram Bot API]
    
    %% === Backend Core ===
    subgraph "Backend Core (FastAPI)"
        API[API Gateway]
        Worker[Background Worker]
        Queue[(Message Queue)]
    end
    
    %% === AI Intelligence Layer ===
    subgraph "AI Intelligence Layer"
        subgraph "Orchestrator Core"
            Orchestrator[Memory-Aware Orchestrator]
            Memory[Mem0 Memory System]
        end
        
        subgraph "Specialist Agents"
            Scheduler[Scheduler Agent]
            Ticketing[Ticketing Agent]
            WebSearch[Web Search Agent]
            Digest[Daily Digest Agent]
        end
        
        subgraph "AI Tools"
            KBSearch[Knowledge Base Search]
            MessageSend[Message Sender]
        end
    end
    
    %% === Data Storage Layer ===
    subgraph "Data Storage Layer"
        SQLite[(SQLite Database)]
        Vector[(ChromaDB Vector Store)]
        Files[File Storage]
    end
    
    %% === External Services ===
    subgraph "External AI & APIs"
        Bedrock[Amazon Bedrock<br/>Claude 3 Haiku]
        GCal[Google Calendar API]
        Tavily[Tavily Search API]
    end
    
    %% === Primary User Flow ===
    Customer -->|sends message| Telegram
    Telegram -->|webhook| API
    API -->|queues| Queue
    Worker -->|processes| Queue
    Worker -->|invokes| Orchestrator
    
    %% === AI Processing Flow ===
    Orchestrator -.->|retrieves context| Memory
    Orchestrator -->|delegates to| Scheduler
    Orchestrator -->|delegates to| Ticketing
    Orchestrator -->|delegates to| WebSearch
    Orchestrator -->|uses| KBSearch
    Orchestrator -->|responds via| MessageSend
    Dashboard -->|delegates to| Digest
    
    %% === Frontend Admin Flow ===
    Dashboard -->|monitoring| API
    Config -->|settings| API
    KB -->|upload docs| API
    Channel -->|manage| API
    
    %% === External Integrations ===
    Scheduler -->|manages events| GCal
    WebSearch -->|real-time search| Tavily
    MessageSend -->|sends replies| Telegram
    
    %% === Data Persistence ===
    Scheduler -.->|stores appointments| SQLite
    Ticketing -.->|stores tickets| SQLite
    Digest -.->|stores reports| SQLite
    Memory -.->|stores conversations| SQLite
    KBSearch -.->|semantic search| Vector
    Files -.->|document storage| SQLite
    
    %% === AI Service Connections ===
    Orchestrator -.->|LLM calls| Bedrock
    Scheduler -.->|LLM calls| Bedrock
    Ticketing -.->|LLM calls| Bedrock
    WebSearch -.->|LLM calls| Bedrock
    Digest -.->|LLM calls| Bedrock
    Memory -.->|embeddings| Bedrock
    KBSearch -.->|embeddings| Bedrock
    
    %% === Styling ===
    classDef user fill:#ffebee,stroke:#d32f2f,stroke-width:3px,color:#000
    classDef frontend fill:#e3f2fd,stroke:#1976d2,stroke-width:2px,color:#000
    classDef backend fill:#fff3e0,stroke:#f57c00,stroke-width:2px,color:#000
    classDef orchestrator fill:#f3e5f5,stroke:#7b1fa2,stroke-width:3px,color:#000
    classDef agents fill:#fff8e1,stroke:#ffa000,stroke-width:2px,color:#000
    classDef tools fill:#e8f5e9,stroke:#388e3c,stroke-width:2px,color:#000
    classDef data fill:#e0f2f1,stroke:#00796b,stroke-width:2px,color:#000
    classDef external fill:#fce4ec,stroke:#c2185b,stroke-width:2px,color:#000
    classDef channel fill:#e1f5fe,stroke:#0277bd,stroke-width:2px,color:#000
    
    class Customer user
    class Dashboard,Config,KB,Channel frontend
    class API,Worker,Queue backend
    class Orchestrator,Memory orchestrator
    class Scheduler,Ticketing,WebSearch,Digest agents
    class KBSearch,MessageSend tools
    class SQLite,Vector,Files data
    class Bedrock,GCal,Tavily external
    class Telegram channel

Architecture Legend:

Customer: End users interacting via Telegram
API Gateway: FastAPI backend handling all requests
Background Worker: Async message processor
Orchestrator: Central AI coordinator with memory
Specialist Agents: Domain-specific AI agents
Knowledge Base: RAG system for company docs
Memory System: Persistent conversation context
Databases: SQLite for structured data, ChromaDB for vectors
Amazon Bedrock: Claude 3 Haiku LLM & Titan embeddings

Technology Stack

Backend:

Framework: FastAPI with async/await support
AI Platform: Amazon Bedrock (Claude 3 Haiku, Titan Embeddings)
Agent Framework: Strands AI for agent orchestration
Memory System: Mem0 for conversation memory and context persistence
Vector Database: ChromaDB for semantic search
Database: SQLite with SQLAlchemy ORM
Document Processing: Docling for PDF parsing
Real-time Search: Tavily API for current information retrieval
Background Processing: Async worker for message queue processing
API Integration: Google Calendar API, Telegram Bot API

Frontend:

Framework: React 18 with functional components
Styling: Tailwind CSS for responsive design
State Management: React hooks (useState, useEffect)
Routing: React Router for SPA navigation
HTTP Client: Fetch API for backend communication

Infrastructure:

Development: Python 3.11+, Node.js, npm
Environment: Virtual environment management
Configuration: Environment variables and database-stored configs
File Handling: Multi-file upload with content type detection
Message Processing: SQLite-based queue with background worker
Memory Storage: Persistent conversation context via Mem0

🎁 Benefits Delivered

For End Customers:

24/7 Availability: Instant responses to inquiries anytime
Intelligent Responses: Context-aware conversations with access to company knowledge
Seamless Scheduling: Easy appointment booking through natural language
Multi-Channel Access: Consistent experience across communication platforms

Technical Benefits:

Modular Architecture: Easy to extend with new agents and channels
Cloud-Native: Leverages AWS services for scalability and reliability
API-First Design: Enables integration with existing business systems
Real-Time Processing: Asynchronous background tasks for optimal performance
Secure Configuration: Database-stored credentials and environment-based settings
Memory-Aware: Persistent conversation context for personalized interactions
Current Information: Real-time web search for up-to-date responses

🛣️ Roadmap and Future Potential

Phase 1 (Current) - Foundation ✅

Multi-agent architecture implementation
Memory-aware orchestrator with Mem0 integration
Telegram integration with background worker processing
Knowledge base RAG system with ChromaDB
Real-time web search capabilities via Tavily
Calendar management with Google Calendar API
Support ticket system with priority management
Business dashboard with daily digest functionality
Native conversations in 15+ languages including English, Chinese, and Spanish

Phase 2 - Enhanced Intelligence & Multi-Channel 🧠

Advanced Sentiment Analysis: Automatic escalation based on customer emotion and urgency
Voice Integration: WhatsApp voice message processing and voice-to-text capabilities
Multi-Modal AI: Process images, documents, and voice messages intelligently
Predictive Customer Service: ML models predicting customer needs and proactive support
Smart Document Processing: Automatic parsing of contracts, invoices, and business documents
Cross-Channel Orchestration: Seamless customer journey across WhatsApp, Email, and Web Chat
Advanced Analytics Dashboard: Real-time insights, customer behavior analysis, and business metrics
Custom Business Intelligence: AI-powered BI dashboards tailored for each company's KPIs and metrics
Parallel Request Handling: Efficiently manage and respond to multiple customer requests at the same time

Phase 3 - Enterprise Integration & AI Workforce 🚀

Enterprise Connectors: Native integration with Salesforce, HubSpot, Slack, and major CRM systems
Custom AI Agent Creation: Build specialized agents for specific business functions
Advanced Workflow Automation: AI-driven business process automation and optimization
Computer Vision Suite: Automatic processing of receipts, IDs, and business documents
Federated Learning: AI that learns across customer interactions while preserving privacy
White-label Solution: Complete platform for businesses to deploy their own AI support system
Advanced Compliance: Industry-specific compliance monitoring (GDPR, HIPAA, financial regulations)
API Marketplace: Extensible platform with third-party integrations and custom plugins
Enterprise BI Suite: Comprehensive business intelligence platform with custom dashboards, automated reporting, and AI-driven insights for executive decision-making

🚀 Getting Started

Prerequisites

Python 3.11+
Node.js 16+
AWS Account (for Bedrock access)
Google Cloud Account (for Calendar API)
Telegram Bot Token
Tavily API Key (for web search functionality)
Mem0 configuration (for memory system)

Quick Setup

Pre-setup

Telegram bot
Google Calender API Client
Expose port 8000

Clone Repository

git clone https://github.com/EwenCheung/SuperConfig.git
cd SuperConfig

Backend Setup

cd backend
python -m venv .venv
source .venv/bin/activate  # On Windows: .venv\Scripts\activate
pip install -r requirements.txt

Environment Configuration

# Create .env file with required credentials
cp .env.example .env
# Edit .env with your API keys

Start Backend

cd ../backend
source .venv/bin/activate
uvicorn src.main:app --reload --host 0.0.0.0 --port 8000

# Open another tab
cd backend
source .venv/bin/activate
python -m src.worker

Frontend Setup

cd frontend
npm install
npm start

Configuration

Access Dashboard: Navigate to http://localhost:3000 in your browser
Configure API Credentials:
- Add your Telegram bot token
- Set up Google Calendar API credentials
- Set tone
Upload Knowledge Base: Upload your company documents (PDFs, docs) for AI training
Customize Settings: Set tone, language preferences, and business-specific configurations
Test Integration: Send a test message via Telegram to verify everything works
🎉 Your personalized AI assistant is ready to serve customers!

Next Steps

Start chatting with your AI agent via Telegram
Monitor customer interactions through the business dashboard
Review daily digest reports for business insights
Scale by adding more knowledge base documents as needed

Name		Name	Last commit message	Last commit date
Latest commit History 93 Commits
.vscode		.vscode
backend		backend
frontend		frontend
.gitignore		.gitignore
README.md		README.md

EwenCheung/SuperConfig

Folders and files

Latest commit

History

Repository files navigation