The complete, production-ready platform for building, deploying, and managing Voice AI Agents.
Built on LiveKit (Real-time WebRTC) and Vobiz (Telephony/SIP).
🚀 An Open Source Alternative to Vapi, Retell AI, and Bland AI.
This platform provides everything you need to build scalable, low-latency voice agents that can handle inbound and outbound calls, manage campaigns, and provide detailed analytics—all in your own infrastructure.
Unlike proprietary APIs that lock you in, this Open Source Voice Agent Platform gives you full control:
- ⚡ Low Latency: Powered by LiveKit's Real-time Transport for instant voice response.
- 📞 Telephony Integration: Seamless SIP trunking via Vobiz for reliable inbound/outbound calls.
- 🤖 Multi-LLM Support: Bring your own models (OpenAI, Deepgram, Azure, Groq).
- 📡 Microservices Architecture: fully modular backend for infinite scalability.
- 📊 Comprehensive Analytics: Built-in call recording, cost tracking, and sentiment analysis.
- 🛠️ Developer First: Complete API, Webhooks, and easy-to-use React frontend.
The platform is designed as a modern microservices system:
graph TD
Client["Frontend / API Client"] --> Gateway["API Gateway (Port 8000)"]
Gateway --> ConfigService["Config Service (Port 8002)"]
Gateway --> AnalyticsService["Analytics Service (Port 8001)"]
Gateway --> OrchestrationService["Orchestration Service (Port 8003)"]
OrchestrationService --> Redis["Redis Queue"]
Redis --> Worker["Agent Worker"]
Worker --> LiveKit["LiveKit Server"]
LiveKit --> Vobiz["Vobiz Telephony"]
Vobiz --> Phone["PSTN Network"]
- Gateway Service: Unified API entry point (Auth, Routing).
- Config Service: Manages Assistants, Phone Numbers, SIP Trunks, and Tools.
- Analytics Service: Handles Call Logs, Recordings (S3), and Analysis.
- Orchestration Service: Manages Outbound Campaigns and Job Queues.
- Agent Worker: Python worker running the actual Voice AI logic (LiveKit Agents).
Get the entire platform running in minutes using Docker.
- Docker & Docker Compose
- Vobiz Account (for SIP Telephony)
- LiveKit Cloud/Self-hosted
- OpenAI/Deepgram API Keys
git clone https://github.com/Piyush-sahoo/Voice-AI-Platform.git
cd Voice-AI-Platform
# Configure backend environment
# Copy example env and fill in your keys
cp backend/.env.example backend/.env.localdocker-compose up -d --buildThis launches:
- Frontend Dashboard:
http://localhost:3000 - API Gateway:
http://localhost:8000 - Agent Worker: Connected to LiveKit
- Redis & MongoDB: Infrastructure
We follow an API-first design. A complete Postman collection is available for testing:
📂 postman/postman_collection.json
POST /api/auth/login- JWT AuthenticationPOST /api/assistants- Create/Manage AI PersonasPOST /api/calls- Trigger Outbound CallsGET /api/analytics/calls- Retrieve Call Logs & RecordingsPOST /api/campaigns- Launch Bulk Call Blasts
We include comprehensive scripts to verify your deployment:
- Full Automation Test: Creates agents, numbers, and makes real calls.
python scripts/full_api_automation.py
- API Key Auth Test: Verifies security and programmatic access.
python scripts/test_api_key_auth.py
├── backend/ # Microservices Backend
│ ├── gateway/ # API Gateway (FastAPI)
│ ├── services/ # Microservices (Config, Analytics, Orchestration)
│ └── shared/ # Shared Libs (Db, Auth, Settings)
├── frontend/ # React/Next.js Dashboard
├── scripts/ # Automation & Testing Scripts
├── postman/ # API Collections
└── docker-compose.yml # Infrastructure Definition
We welcome contributions! Please see our Contribution Guidelines for a step-by-step guide on how to:
- Fork and Clone the repo
- Set up your development environment
- Create a Pull Request (PR)
- Follow our coding standards
- Issues: Submit bugs/feature requests on GitHub.
- Discussions: Ask questions and share ideas.