Dograh

What is Dograh?

Dograh is an open-source platform for building AI voice agents that make and answer phone calls. Teams design call flows in a visual builder, then pick their own speech-to-text, LLM, text-to-speech, and telephony providers. You can self-host it for free, use the managed cloud, or have the team deploy it inside your private cloud.

Top Features:

• Visual workflow builder: map agent nodes, tools, transitions, and call outcomes on one canvas.
• Bring your own models: swap providers like OpenAI, Deepgram, and ElevenLabs without vendor lock-in.
• MCP server: coding agents like Claude Code can create and edit voice agents.

Use Cases:

• Lead qualification: call new form submissions and screen them for purchase intent.
• Payment reminders: run EMI reminder and soft collection calls at scale.
• Support lines: answer inbound calls and hand off to humans when needed.

Who Can Use Dograh?

• Developers: build and test voice bots with Python and Node SDKs.
• Regulated companies: keep calls, recordings, and transcripts inside their own perimeter.
• Startups: replace closed voice platforms like Vapi and Retell with open code.

Pricing

• Open Source (free): self-host the full platform with no usage limits under BSD license.
• Pay as You Go ($0.01/min): platform fee on the managed cloud with 10 concurrent calls.
• Enterprise (contact sales): on-premise deployment, SSO, audit logs, and custom SLAs.

Pros and Cons

Pros:

• No lock-in: every part of the voice stack can be swapped or forked.
• Data control: on-prem and air-gapped setups keep recordings away from third parties.
• Low platform fee: the cloud plan charges one cent per minute before model costs.

Cons:

• Technical setup: self-hosting needs Docker skills and someone to run the servers.
• Separate model bills: bring-your-own-key setups add separate provider costs on top.
• Beta pieces: swapping self-hosted LLMs is still in beta for now.

FAQs:

1) Is Dograh really open source?

Yes, the full platform is released under the BSD 2-Clause license on GitHub.

2) How is it different from Vapi or Retell?

It is open source, self-hostable, and lets you swap every provider.

3) Which phone providers does it support?

It connects to Twilio, Vonage, Telnyx, Plivo, Cloudonix, Asterisk, and several others.

4) Does it support speech-to-speech models?

Yes, it can use Gemini 3.1 Flash Live or GPT Realtime 2 for lower latency.

5) Can I try it without hosting it?

Yes, the managed cloud lets you test agents in the browser before going live.