My Workflow 4¶

A multilingual AI assistant workflow that processes both text and voice messages through WhatsApp, providing intelligent responses with translation capabilities between English, Luganda, and Swahili.

Purpose¶

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

This workflow appears to be designed for multilingual communication support, enabling users to interact with an AI assistant through WhatsApp in multiple languages. The system can handle both text messages and voice notes, automatically transcribing audio and providing translation services between English and local African languages (Luganda and Swahili).

How It Works¶

The workflow operates through two main pathways:

Chat Interface Path: 1. Receives chat messages through a chat trigger 2. Simultaneously translates the input to Luganda and processes it through an AI agent 3. The AI agent uses OpenAI's GPT-4 model with translation tools and memory to provide contextual responses

WhatsApp Integration Path (Currently Disabled): 1. Receives WhatsApp messages via webhook 2. Checks if the message contains audio content 3. For voice messages: downloads the audio file, transcribes it using OpenAI's Whisper, then processes the text 4. For text messages: directly processes the message content 5. Translates the final message to Luganda and sends it to the AI agent for response generation

Both paths utilize conversation memory to maintain context across interactions and can translate between English, Luganda, and Swahili as needed.

Workflow Diagram¶

graph TD
    A[When chat message received] --> B[Translate a language]
    A --> C[AI Agent]

    D[WhatsApp Webhook] --> E[If Voicenote1]
    E -->|Audio| F[Download Voice File1]
    E -->|Text| G[Text In]

    F --> H[Transcribe a recording1]
    H --> I[Voice Out]
    I --> J[User Message]
    G --> J

    J --> K[Translate a language1]
    J --> L[AI Agent1]

    M[Simple Memory] -.->|Memory| C
    N[Translate a language in Google Translate] -.->|Tool| C
    O[OpenAI Chat Model] -.->|LLM| C

    P[Simple Memory1] -.->|Memory| L
    Q[Translate a language in Google Translate1] -.->|Tool| L
    R[OpenAI Chat Model1] -.->|LLM| L

    style D fill:#ffcccc
    style E fill:#ffcccc
    style F fill:#ffcccc
    style G fill:#ffcccc
    style H fill:#ffcccc
    style I fill:#ffcccc
    style J fill:#ffcccc
    style K fill:#ffcccc
    style L fill:#ffcccc
    style P fill:#ffcccc
    style Q fill:#ffcccc
    style R fill:#ffcccc

Trigger¶

Primary Trigger: Chat Trigger - Responds to incoming chat messages
Secondary Trigger: WhatsApp Webhook (currently disabled) - Receives WhatsApp messages via HTTP POST to /coaching-start endpoint

Nodes Used¶

Node Type	Purpose
Chat Trigger	Receives and processes incoming chat messages
WhatsApp Webhook	Handles incoming WhatsApp messages (disabled)
Google Translate	Translates text to Luganda language
Google Translate Tool	Provides translation capabilities as an AI tool
AI Agent	Processes messages using OpenAI with translation system prompt
OpenAI Chat Model	GPT-4 language model for generating responses
Simple Memory	Maintains conversation context and history
OpenAI Transcribe	Converts voice messages to text using Whisper
HTTP Request	Downloads voice files from WhatsApp
If Node	Conditional logic to handle voice vs text messages
Set Node	Data transformation and message formatting

External Services & Credentials Required¶

Required Credentials:¶

Google Service Account: For Google Translate API access
OpenAI API: For GPT-4 chat model and Whisper transcription

External Services:¶

Google Translate API: Text translation between languages
OpenAI API: Language model and speech-to-text services
WhatsApp Business API (via Twilio): Message receiving and sending
Twilio Webhook: For WhatsApp integration

Environment Variables¶

No specific environment variables are defined in this workflow. All configuration is handled through n8n credentials and node parameters.

Data Flow¶

Input:¶

Chat messages: Text input through chat interface
WhatsApp messages: Text or voice messages via webhook
Voice files: Audio files from WhatsApp media URLs

Processing:¶

Text translation to/from Luganda and Swahili
Voice transcription to text
AI-powered response generation with conversation memory
Language detection and appropriate translation

Output:¶

AI responses: Contextual replies in appropriate language
Translated text: Messages converted between supported languages
Transcribed text: Voice messages converted to text format

Error Handling¶

The workflow includes basic error handling through: - Conditional logic for message type detection (voice vs text) - alwaysOutputData setting on AI agents to ensure execution continues - Separate processing paths to isolate failures

No explicit error handling nodes or try-catch mechanisms are implemented.

Known Limitations¶

WhatsApp integration pathway is currently disabled
Limited to three languages: English, Luganda, and Swahili
No explicit error handling for API failures
Voice file processing depends on WhatsApp media URL accessibility
Memory is session-based and may not persist across workflow restarts

No related workflows specified in the current context.

Setup Instructions¶

1. Import Workflow¶

Copy the workflow JSON
In n8n, go to Workflows → Import from JSON
Paste the JSON and import

2. Configure Credentials¶

Google Service Account:
- Create a Google Cloud project
- Enable Google Translate API
- Create a service account and download JSON key
- Add credential in n8n with the service account details
OpenAI API:
- Obtain OpenAI API key
- Add credential in n8n with your API key
- Ensure access to GPT-4 and Whisper models

3. Configure Webhooks (Optional)¶

For WhatsApp integration:
- Set up Twilio WhatsApp Business account
- Configure webhook URL: https://your-n8n-instance.com/webhook/coaching-start
- Enable the disabled WhatsApp webhook node

4. Test the Workflow¶

Activate the workflow
Test the chat trigger with sample messages
Verify translation functionality
Test voice transcription if WhatsApp integration is enabled

5. Customize System Prompts¶

Modify the AI Agent system messages to match your specific use case
Adjust translation target languages as needed
Configure memory settings based on conversation length requirements