Business Voice Agent - Backend

• Live Example: AI Agent Dave
• Frontend Repository: business_voice_agent_frontend
• Interest Form Share your interest by completing this form

A voice-enabled AI assistant backend for David McGrath's business website (digispectintelligence.com). This backend powers Agent Dave, providing real-time voice interaction capabilities using Restack AI.

Overview

Agent Dave is designed to assist visitors of Digispect Intelligence by providing information about David's profile, expertise, and services. This backend handles the voice processing pipeline, AI model integration, and real-time communication.

Prerequisites

• Docker (for running Restack)
• Python 3.10 or higher
• Restack (docker image for local development and account for web hosting)
• Livekit account (for live chat)
• Deepgram account (for speech-to-text transcription)
• ElevenLabs account (for text-to-speech and voice cloning)
• OpenAI (for AI logic)

Configuration

1. Copy the .env.example file in the agent subfolder and the .env.local.example file in the livekit_pipeline subfolder (and rename by removing the .example suffixes):

2. Configure the following services and update the values in the above mentioned files:

   Livekit Setup

   • Sign up at Livekit
   • In your .env and .env.local files:
     • Set LIVEKIT_URL to WEBSOCKET_URL value
     • Set LIVEKIT_API_KEY to API_KEY value
     • Set LIVEKIT_API_SECRET to API_SECRET value

   Deepgram Setup

   • Sign up at Deepgram
   • Obtain an API key and add to DEEPGRAM_API_KEY in .env.local file

   ElevenLabs Setup

   • Sign up at ElevenLabs
   • Add ELEVEN_API_KEY and ELEVENLABS_VOICE_IDto .env.local file

   OpenAI Setup

   • Sign up at OpenAI
   • Add OPENAI_API_KEYto .env file

Installation and Setup

Note: Make sure to also set up the business_voice_agent_frontend repository for the complete system.

1. Start Restack

docker run -d --pull always --name restack -p 5233:5233 -p 6233:6233 -p 7233:7233 -p 9233:9233 -p 10233:10233 ghcr.io/restackio/restack:main

2. Start Restack Agent with Voice

Create Python environment

Using uv:

cd agent
uv venv && source .venv/bin/activate
uv sync
uv run dev

Using pip:

cd agent
python -m venv .venv && source .venv/bin/activate
pip install -e .
python -c "from src.services import watch_services; watch_services()"

3. Start Livekit Voice Pipeline

Create Python environment

Using uv:

cd livekit_pipeline
uv venv && source .venv/bin/activate
uv sync
uv run python src/pipeline.py dev

Using pip:

cd livekit_pipeline
python -m venv .venv && source .venv/bin/activate
pip install -e .
python src/pipeline.py dev

Usage

This backend works in conjunction with the business_voice_agent_frontend. Please refer to the frontend repository for complete usage instructions on how to interact with the voice assistant system.

Deployment

To deploy on Restack Cloud:

1. Create an account at https://console.restack.io
2. Follow the deployment instructions in the Restack documentation

License

This project is licensed under the MIT License - see the LICENSE file for details.

Contact

For questions or support, contact david.mcgrath@digispectintelligence.com

References

This repo is based on the examples here:

• https://github.com/AnswerDotAI/fasthtml-example
• https://github.com/restackio/examples-python/tree/main/agent_voice/livekit
• https://github.com/livekit/agents/tree/main/examples/voice_agents
• https://github.com/livekit/client-sdk-js/tree/main/examples