Dograh AI

Dograh AI is an open-source voice AI platform built by Zansat Technologies Private Limited, founded by YC alumni and exit founders. It offers a self-hostable alternative to proprietary platforms like Vapi and Retell, letting teams build and deploy production voice agents with full control over their infrastructure and data. The project is licensed under BSD 2-Clause and is actively maintained on GitHub, with a cloud version also available at app.dograh.com.

What It Is

Dograh AI is a voice agent platform that combines a visual drag-and-drop workflow builder with support for inbound and outbound telephony, custom LLM/TTS/STT integrations, and MCP-native capabilities. It is designed to get teams from zero to a working voice bot in under two minutes, with no API keys required out of the box — Dograh ships with auto-generated keys and its own LLM, TTS, and STT stack. Users can connect their own providers (such as OpenAI, Twilio, Vonage, or Telnyx) at any time.

Self-Hosting and Deployment Model

Dograh is designed to be deployed with a single Docker command, making it accessible for local development and remote server deployments alike. Key deployment options include:

• Local: One docker-compose command launches the full stack at localhost:3010
• Remote server: HTTPS-enabled remote deployment documented in the official Docker Deployment Guide
• Cloud: A managed cloud version is available at app.dograh.com for teams that prefer not to self-host

The platform is Python-based and containerized, with a modular architecture that allows swapping components as needed. Anonymous telemetry is collected by default but can be disabled via an environment variable.

Voice Capabilities and Integrations

Dograh supports a range of telephony providers out of the box, including Twilio, Vonage, Vobiz, and Cloudonix, with the ability to add others. It supports call transfers to human agents and real-time, low-latency voice interactions. Language support ships with English and is described as expandable. Developers can bring their own TTS and STT models, or use Dograh's built-in stack.

The platform is MCP-native and includes a built-in QA node that analyzes prompt quality across workflow nodes. In-dashboard web calls allow developers to test agents directly while building, without any telephony setup.

Developer Experience and SDKs

Dograh provides both a Python SDK (available on PyPI as dograh-sdk) and a Node.js SDK (available on npm as @dograh/sdk). The platform is Python-based internally, making it straightforward to customize. A test mode lets developers run end-to-end agent tests without affecting production calls or data.

Community support is centered on a Slack workspace and GitHub Discussions, with the founders described as personally onboarding early adopters.

Update: v1.32.0

The latest release is dograh-v1.32.0, published on May 28, 2026. The repository was created in September 2025 and has seen rapid iteration, with the last push recorded in late May 2026. The project has accumulated over 3,800 GitHub stars and 779 forks since launch, and is listed on Trendshift. The repository's active issue count and release cadence signal ongoing, production-focused development.