Modular voice stack
Dograh lets you assemble a voice agent from separate inbound, speech-to-text, language model, text-to-speech, and telephony components, or switch to a single speech-to-speech pipeline.
Dograh
Dograh is open-source voice agent infrastructure for building and running voice agents on your own servers, in your VPC, or in Dograh’s managed cloud. It combines modular voice pipelines, MCP-based agent building, and a speech-to-speech option for real-time conversations.
Open-source infrastructure for building voice agents
Dograh is open-source voice agent infrastructure for teams that want to build, deploy, and run voice agents on infrastructure they control. The product is positioned as a self-hostable alternative, with options to run on your own servers, inside your VPC, or through Dograh’s managed cloud.
The site shows two main ways to build agents: a modular cascade with inbound channel, STT, LLM, TTS, and telephony components, or a speech-to-speech flow that keeps audio in and audio out without a text round trip. Dograh also ships a Model Context Protocol server so agentic IDE tools can create and modify agents directly against the stack.
The platform is aimed at teams that care about data residency, auditability, and deployment flexibility. Dograh states that self-hosted or private-cloud deployments can keep calls, recordings, transcripts, prompts, customer PII, and even model inference within the customer boundary, and the pricing page says the open-source platform is free forever when self-hosted.
Core capabilities
Multiple deployment modes
The product is built around deployment control: run it on your own servers, inside your VPC, or in Dograh’s managed cloud, depending on how much operational responsibility you want to keep.
MCP-based agent building
Dograh ships a Model Context Protocol server so agentic tools can create, edit, and deploy voice agents directly from the IDE or other MCP clients.
Speech-to-speech workflow
The site describes support for real-time audio-to-audio conversations, including turn-taking and interruption handling, with Gemini 3.1 Flash Live or GPT Realtime 2 mentioned for speech-to-speech use.
Bring-your-own model stack
For self-hosted deployments, Dograh says teams can plug in models that run in their own perimeter, including STT and TTS options such as Whisper, Voxtral, Canary Qwen, Kokoro, and Chatterbox.
Usage-based self-serve pricing
The pricing page states that the self-serve plan includes BYOK or Dograh models, and that the platform supports usage-based billing with 10 concurrent calls included on the self-serve tier.
Practical ways teams use Dograh
Deploy inside a controlled data perimeter
Use Dograh when you need a voice agent platform that can live inside your own infrastructure boundary, whether that means on-premise servers, a private cloud, or a VPC you control.
Build agents from MCP-compatible development tools
Use the MCP server to build or update agents from an IDE when your team prefers agentic coding workflows and wants to generate call scripts, nodes, and deployment wiring without switching tools.
Run real-time audio conversations
Choose the speech-to-speech path for live conversations that benefit from lower latency and more natural turn-taking, such as interruption-prone calls or interactive assistants.
Compose a custom voice stack
Use the modular stack when you want to swap individual parts of the pipeline, such as changing telephony, STT, LLM, or TTS providers without rebuilding the whole system.
Support regulated-industry voice workflows
Adopt the platform for regulated use cases where data residency, auditability, and self-hosting are important, such as fintech, healthtech, legal intake, insurance, banking, or government workflows.
Pros and Cons
Pros
- Open-source and self-hostable, with the option to run the full platform on your own infrastructure.
- Supports multiple deployment models, including self-hosted, managed cloud, and private-cloud/VPC deployment.
- Can be driven from MCP-compatible tools such as Claude Code and Cursor, which may fit agentic development workflows.
- Offers both modular cascade and speech-to-speech agent flows, giving teams a choice between control and lower-latency audio conversations.
- Pricing information is clear at a high level: self-hosted is free forever, and managed usage is pay-as-you-go with custom volume and enterprise options.
Cons
- The reviewed pages do not provide detailed documentation for every supported integration or model provider, so some capabilities remain unclear from the site alone.
- LLM swapping is described as being in beta for self-hosted deployments, which suggests parts of the stack are still maturing.
FAQ
Can Dograh be self-hosted or run in a managed environment?
Dograh is designed to run either on your own infrastructure or in Dograh’s managed cloud. The site also describes a private-cloud deployment where Dograh deploys the stack inside your VPC and operates it for you.
How does the MCP workflow work?
The homepage says Dograh ships a Model Context Protocol server, so tools such as Claude Code, Cursor, OpenClaw, Codex, or other MCP clients can connect to it and create or modify voice agents from the IDE.
What kind of voice agent pipeline does Dograh support?
The site presents a cascade pipeline with inbound channels, STT, LLM, TTS, and telephony, and it also describes a speech-to-speech option for real-time audio in and audio out.
Is there a free self-hosted option and a paid plan?
Dograh’s pricing page says the open-source platform is free forever to self-host, while its managed cloud uses pay-as-you-go pricing starting at 1¢ per minute platform fee, with custom volume and enterprise options available.
What kinds of teams or workflows does Dograh appear to fit best?
The site highlights regulated-industry and jurisdictional use, including fintech, healthtech, telemedicine, insurance, banking, legal intake, pharma, defense, and government, but it does not provide detailed setup guidance or product limits on the pages reviewed.
Quick Facts
- Category
- Open-source voice agent infrastructure
- Primary deployment
- Self-hosted, managed cloud, or private cloud/VPC
- Primary users
- Teams building voice agents, especially in regulated environments
- Source domain
- dograh.com
- Build workflow
- MCP-driven agent creation plus modular voice pipelines
- Pricing model
- Free self-hosted OSS; paid managed usage-based plans
Alternatives à Dograh
CreateOS Sandbox
CreateOS Sandbox is an isolated compute environment for running code and agent workloads inside Firecracker micro-VMs. It is designed for workflows that need machine-level isolation, private networking between sandboxes, and programmatic control through SDK, CLI, or MCP.
Wallie
Wallie is an open-source AI streamer that watches your screen, hears chat, and generates live commentary in a configurable persona. It runs locally on your machine with your own keys and is aimed at faceless content, autonomous streams, and real-time reactions.
AakarDev AI
AakarDev AI helps teams manage AI provider access, project-level setups, logs, and analytics from one dashboard. It supports BYOK workflows and lists providers including OpenAI, Google Gemini, Anthropic, Groq, Mistral AI, and Perplexity AI.
Trigger.dev chat agent
Trigger.dev chat agent is a durable AI chat backend for developers building stateful conversations that can survive refreshes, crashes, and long-running turns. It connects with the AI SDK `useChat` flow and runs on managed infrastructure with no timeout on a turn.
ByteAsk
ByteAsk is a terminal-first AI coding agent for C and C++ that edits repositories and verifies changes with the real compiler, debugger, sanitizers, and tests before showing a diff. It offers a free tier plus paid plans, with editor connectors and zero-retention handling described in the source.
Codex Plugins
Codex Plugins bundle reusable skills, app integrations, and MCP servers into workflows you can install in the Codex app or use from Codex CLI. They help extend Codex with connected-service tasks, reusable instructions, and shared team workflows.