A self-hostable voice AI platform providing a visual builder for voice agents with an integrated LLM, TTS, and STT stack for immediate testing. It supports optional telephony integrations and Bring Your Own Model (BYOK) configurations for deployment without proprietary SaaS lock-in.
Project overview
Enables source-level control and on-premises data residency for voice agents, offering a native AI stack that allows immediate interaction without external API keys.
Developers and teams seeking a self-hostable voice AI platform with source-level customization.
Key capabilities
Enables users to visually create and configure production voice agents quickly without coding.
Ships with an integrated LLM, TTS, and STT stack for immediate voice interactions via web calls without initially needing external API keys.
Includes native integration with MCP (Model Context Protocol).
Allows users to connect their own custom TTS, STT, and LLM models or provider keys.
Limitations and risks
Collects anonymous usage data by default unless explicitly disabled via an environment variable.
Getting started
Download docker-compose.yaml and start_docker.sh via curl, execute ./start_docker.sh, open http://localhost:3010, and click Web Call to interact with the bot.
Alternatives and comparisons
Provides a no-code visual builder and a TypeScript SDK for creating and managing AI agents.
An open-source Agent chatbot platform that integrates with mainstream IM apps and supports plugins and LLMs.
A desktop app and local runtime for managing multi-platform AI agent chats, sessions, and models.
README: **The open-source, self-hostable alternative to Vapi & Retell** — build production voice agents with a drag-and-drop workflow builder. From zero to a working bot in under 2 minute…
GitHub project description: On Prem, BYOK across Speech to Speech or LLM/STT/TTS, with a visual workflow builder, MCP native and telephony support.
README: | **Self-hostable** | ✅ Yes — one Docker command | ❌ SaaS only | ❌ SaaS only |
README: We collect anonymous usage data to improve the product. You can opt out by setting `ENABLE_TELEMETRY=false` before running the startup script.
README: Click **Web Call** — you're talking to your bot.