Vogent Inbound Call Transcript Automation

This Python application automatically retrieves and saves transcripts from Vogent inbound calls. It runs a Flask webhook server that receives notifications from Vogent when calls complete, then fetches and saves the transcripts without any manual intervention.

Features

• ✅ Fully Automated: No manual steps required after setup
• ✅ Real-time Processing: Uses Vogent webhooks for immediate transcript retrieval
• ✅ Readable Output: Saves transcripts in human-readable format
• ✅ Backup Data: Also saves raw JSON data for debugging
• ✅ Error Handling: Comprehensive logging and error recovery
• ✅ Manual Override: API endpoint for manual transcript retrieval
• ✅ Health Monitoring: Built-in health check endpoint

How It Works

1. Inbound Call: Someone calls your Vogent-linked phone number
2. AI Agent Handles Call: Your configured Vogent agent answers and conducts the conversation
3. Webhook Notification: Vogent sends a dial.transcript webhook when the transcript is ready
4. Automatic Retrieval: The script fetches the transcript via Vogent API
5. File Saving: Transcript is saved as a formatted text file in the transcripts/ directory

Quick Start

1. Install Dependencies

# Install Python packages
pip install -r requirements.txt

2. Configure Environment

# Copy the example environment file
copy .env.example .env

Edit .env and add your Vogent API key:

VOGENT_API_KEY=elto_YourActualAPIKeyHere

3. Configure Vogent Webhooks

In your Vogent dashboard:

1. Go to API Settings
2. Set Webhook URL to: http://your-server:5000/webhook/vogent
3. Enable the dial.transcript event

Note: For testing locally, use a tool like ngrok to expose your local server:

# In another terminal
ngrok http 5000
# Use the ngrok URL as your webhook URL in Vogent

4. Run the Application

python vogent_transcript_automation.py

The server will start on port 5000 and begin listening for webhooks.

Usage

Automatic Operation

Once running, the system operates completely automatically:

1. When someone calls your Vogent number, the AI agent handles the call
2. After the call ends and transcript is ready, Vogent sends a webhook
3. The script automatically fetches and saves the transcript
4. Files are saved in the transcripts/ directory with timestamps

Manual Transcript Retrieval

You can also manually fetch a transcript if needed:

# Using curl or similar HTTP client
curl -X POST http://localhost:5000/manual-fetch/YOUR_DIAL_ID

Monitoring

Check if the service is running:

curl http://localhost:5000/health

File Output

The system creates two files for each call:

1. Formatted Transcript (transcript_[dial_id]_[timestamp].txt)

============================================================
VOGENT CALL TRANSCRIPT
============================================================
Dial ID: 5a6c6190-db20-4d8e-86a9-79a6af292dea
Generated: 2025-10-05 14:30:22
Agent ID: 97512a3c-dd94-45fb-965c-e8d58e762fcb
Caller Number: +18001234567
Call Status: completed

------------------------------------------------------------
CONVERSATION TRANSCRIPT
------------------------------------------------------------

[001] AI: Hello, thank you for calling Dr. Smith's office. How can I help you today?

[002] HUMAN: Hi, I need to schedule an appointment for a check-up.

[003] AI: I'd be happy to help you schedule that appointment. Could you please provide me with your full name?

[004] HUMAN: Sure, it's John Doe.

------------------------------------------------------------
End of transcript - 4 total entries

2. Raw JSON Data (raw_dial_[dial_id]_[timestamp].json)

Contains the complete API response from Vogent for debugging purposes.

Configuration Options

Environment Variables

Variable	Required	Default	Description
VOGENT_API_KEY	Yes	-	Your Vogent API key (starts with elto_)
PORT	No	5000	Port for the webhook server
DEBUG	No	False	Enable Flask debug mode

Directory Structure

Inbound_transcript/
├── vogent_transcript_automation.py  # Main application
├── requirements.txt                 # Python dependencies
├── .env                            # Your configuration (create from .env.example)
├── .env.example                    # Configuration template
├── transcripts/                    # Output directory (auto-created)
│   ├── transcript_[id]_[time].txt  # Formatted transcripts
│   └── raw_dial_[id]_[time].json   # Raw API responses
└── vogent_automation.log           # Application logs

API Endpoints

Endpoint	Method	Description
/webhook/vogent	POST	Receives Vogent webhooks
/health	GET	Health check
/manual-fetch/<dial_id>	POST	Manually fetch a transcript

Troubleshooting

Common Issues

1. "VOGENT_API_KEY environment variable not set!"

• Solution: Make sure you've created a .env file with your API key

2. Webhooks not being received

• Check that your webhook URL is correctly configured in Vogent
• Ensure your server is accessible from the internet (use ngrok for local testing)
• Verify the webhook URL includes the correct path: /webhook/vogent

3. Transcript not found

• The transcript might not be ready immediately after the call
• The script includes automatic retry logic
• Check the logs for specific error messages

4. Connection errors

• Verify your internet connection
• Check if the Vogent API is accessible: https://api.vogent.ai/api
• Ensure your API key is valid and has the correct permissions

Logging

The application logs to both:

• Console output (visible when running)
• vogent_automation.log file

Log levels:

• INFO: Normal operations (calls received, transcripts saved)
• WARNING: Non-critical issues (transcript not immediately ready)
• ERROR: Critical problems (API failures, file write errors)

Testing

You can test the webhook endpoint manually:

# Test dial.transcript webhook
curl -X POST http://localhost:5000/webhook/vogent \
  -H "Content-Type: application/json" \
  -d '{
    "event": "dial.transcript",
    "payload": {
      "dial_id": "test-dial-id-123"
    }
  }'

Production Deployment

For production use:

1. Use a proper WSGI server (not Flask's dev server):

   pip install gunicorn
   gunicorn -w 4 -b 0.0.0.0:5000 vogent_transcript_automation:app

2. Set up reverse proxy (nginx/Apache) for HTTPS

3. Configure proper logging and log rotation

4. Set up monitoring and alerting

5. Secure your server and restrict access to webhook endpoints

Security Considerations

• Keep your Vogent API key secure and never commit it to version control
• Use HTTPS for webhook URLs in production
• Consider implementing webhook signature verification
• Restrict server access and use firewalls appropriately
• Regularly rotate API keys and monitor for unauthorized access

Support

For issues with:

• This script: Check the logs and troubleshooting section above
• Vogent API: Refer to Vogent documentation
• Webhook configuration: Check your Vogent dashboard settings

License

This automation script is provided as-is for use with the Vogent platform.