Work_Flow_Of_Ali

A multilingual Telegram customer service bot that provides intelligent support in both English and Luganda, using AI-powered language detection, knowledge base search, and conversation memory to deliver personalized customer assistance.

Purpose

No business context provided yet — add a context.md to enrich this documentation.

How It Works

1. Message Reception: The workflow activates when a user sends a message to the connected Telegram bot
2. User Profile Lookup: Retrieves the user's existing profile from the PostgreSQL database using their Telegram ID
3. Language Detection: Analyzes the incoming message text to detect whether it's written in English or Luganda using keyword matching
4. AI-Powered Response: An OpenAI-powered agent processes the message, searches the knowledge base for relevant information, and generates a response in the same language as the user's message
5. Profile Update: Updates the user's language preference in the database based on the detected language
6. Response Delivery: Sends the AI-generated response back to the user via Telegram

The system maintains conversation history and personalizes responses based on previous interactions, ensuring continuity across multiple conversations.

Mermaid Diagram

graph TD
    A[Telegram Trigger] --> B[Get User Profile]
    B --> C[Detect Language]
    C --> D[Customer Service Agent]
    E[OpenAI Chat Model] --> D
    F[Postgres Chat Memory] --> D
    G[Knowledge Base Vector Store] --> D
    H[HuggingFace Embeddings] --> G
    D --> I[Update User Profile]
    I --> J[Send Telegram Response]

Trigger

Telegram Trigger: Webhook-based trigger that activates when users send messages to the configured Telegram bot. Listens for message updates and captures the full message context including sender information.

Nodes Used

Node Type	Purpose
Telegram Trigger	Receives incoming messages from Telegram users
Postgres (Get User Profile)	Retrieves existing user data from the database
Code (Detect Language)	Analyzes message text to identify English vs Luganda language
OpenAI Chat Model	Provides the AI language model for generating responses
Postgres Chat Memory	Stores and retrieves conversation history for context
Knowledge Base Vector Store	Searches the knowledge base for relevant information
HuggingFace Embeddings	Converts text to vector embeddings for semantic search
Customer Service Agent	Orchestrates the AI response generation with tools and memory
Postgres (Update User Profile)	Updates user language preferences in the database
Telegram (Send Response)	Delivers the AI-generated response back to the user

External Services & Credentials Required

• Telegram Bot API: Bot token for receiving and sending messages
• OpenAI API: API key for GPT-4.1-mini language model access
• HuggingFace API: API token for text embedding services
• PostgreSQL Database: Connection credentials for user profiles, chat history, and knowledge base storage

Environment Variables

The workflow requires the following credentials to be configured in n8n: - Telegram bot credentials (bot token) - OpenAI API credentials - HuggingFace API credentials - PostgreSQL database connection details

Data Flow

Input: Telegram messages containing user text in English or Luganda

Processing: - User identification and profile retrieval - Language detection using keyword analysis - AI-powered response generation with knowledge base search - Conversation history integration - User profile updates

Output: - Contextual responses in the user's preferred language - Updated user profiles with language preferences - Stored conversation history for future interactions

Error Handling

The workflow does not include explicit error handling nodes. Consider adding error handling for: - Database connection failures - API rate limits or timeouts - Invalid Telegram message formats - Knowledge base search failures

Known Limitations

No specific limitations documented in the provided context.

Related Workflows

No related workflows mentioned in the provided context.

Setup Instructions

1. Import Workflow: Import the JSON into your n8n instance

2. Configure Credentials:

   • Set up Telegram bot credentials with your bot token
   • Configure OpenAI API credentials with your API key
   • Add HuggingFace API credentials
   • Set up PostgreSQL database connection

3. Database Setup:

   -- Create users table
   CREATE TABLE users (
     telegram_id BIGINT PRIMARY KEY,
     language VARCHAR(10),
     created_at TIMESTAMP DEFAULT CURRENT_TIMESTAMP
   );
   
   -- Create chat_history table for conversation memory
   CREATE TABLE chat_history (
     session_id VARCHAR(255),
     message_id SERIAL,
     message JSONB,
     created_at TIMESTAMP DEFAULT CURRENT_TIMESTAMP
   );
   

4. Knowledge Base Setup:

   • Populate the vector store with your knowledge base content
   • Ensure the HuggingFace embeddings are properly configured

5. Telegram Bot Configuration:

   • Create a bot via @BotFather on Telegram
   • Set the webhook URL to your n8n instance
   • Configure the bot's privacy settings as needed

6. Test the Workflow:

   • Activate the workflow in n8n
   • Send test messages in both English and Luganda to verify language detection
   • Confirm responses are generated in the correct language
   • Verify user profiles are being created and updated

7. Monitor and Optimize:

   • Review conversation logs for accuracy
   • Update the knowledge base as needed
   • Adjust language detection keywords for better accuracy