Dograh
Open-source, self-hostable builder for voice AI agents — an alternative to Vapi and Retell.
Dograh is an open-source visual workflow builder for voice agents, letting you swap the speech-to-text, LLM, text-to-speech, and telephony provider at each node, or skip the cascade entirely for a single speech-to-speech model. It's built to self-host on-prem or inside your own VPC, with an option to run models fully air-gapped so no call audio, transcripts, or PII leave your boundary, aimed at teams that can't send calls to a cloud vendor. It also ships an MCP server, so Claude Code or Cursor can build and deploy a voice agent by prompt without leaving the IDE.
What holds up
- +Self-hostable on-prem or in your own VPC, with an option to run models fully air-gapped
- +Swap the STT, LLM, TTS, and telephony provider per node instead of locking into one vendor's stack
- +Ships an MCP server, so Claude Code or Cursor can build and deploy an agent directly from the IDE
Mind the limits
- −Self-hosting is on you — running on-prem or in your own VPC means owning the infrastructure, not a hosted SaaS
- −Compliance framing (HIPAA, GDPR, SOC 2, ISO 27001) describes the architecture, not a stated certification
Featured here? Take the badge
Put it on your site — it links back to this review. Free for every listed product, always.
<a href="https://stillworks.dev/products/p/dograh/"><img src="https://stillworks.dev/badge/dograh.svg" alt="Picked by StillWorks" width="250" height="54"></a>