Bolna
An MIT voice-agent platform that defines an agent in a JSON file and orchestrates speech to text, a language model and text to speech over websockets, with Twilio and Plivo for phone calls, whose maintainers say they are looking for help.
What it does
Bolna is an "end-to-end open-source voice agents platform" that lets you "quickly build voice firsts conversational assistants through a json", in its README's words. It "orchestrates voice conversations using combination of different ASR+LLM+TTS providers and models over websockets".
Telephony is Twilio and Plivo, with Exotel and Vonage marked coming soon. Transcription names Deepgram and Azure. Language models go through LiteLLM, so OpenAI, DeepSeek, Llama, Cohere, Mistral and others work, plus models served by vLLM. Text to speech lists AWS Polly, ElevenLabs, Deepgram, OpenAI, Azure, Cartesia and others. A text-only pipeline with no speech is also supported.
What it replaces
Bolna replaces Vapi and Retell AI for a phone agent defined in a file and run in four containers, where the models are hosted services you already have keys for.
Where it falls short, in the maintainers' own words: "We are actively looking for maintainers" and "We have fluctuated b/w maintaining this repository purely from a point of time crunch and not interest." The local example setup is marked for a move to another repository. Interruption handling and WebRTC are not described. The maintainers sell hosted APIs and a no-code playground built on this code, both closed source.
How it runs
Docker Compose from the local setup folder, starting a telephony web server (Twilio or Plivo), the Bolna server, ngrok and Redis. It can also be imported "directly in Python without the local telephony setup". A Twilio or Plivo account and an ngrok token are needed for calls. The code is MIT with no added conditions.
Who it fits
A small team that wants a phone agent running today from hosted providers, and is comfortable with a project whose maintenance its own README describes as uneven. For a larger build, Pipecat or LiveKit Agents.
Limits the README states
The maintainer capacity notes above; the pending move of the local setup; Exotel and Vonage not yet available.
Sources
- Bolna README and MIT LICENSE: github.com/bolna-ai/bolna, read 2026-09-16.
More voice agent frameworks
A Python framework under BSD-2-Clause, maintained by Daily, for real-time voice and multimodal agents, with the widest provider list on this site and telephony serialisers for Twilio, Telnyx, Plivo, Vonage, Exotel and Genesys.
An Apache-2.0 framework for real-time voice agents that connects any speech, language and voice provider, with semantic turn detection, native MCP tool support, a test framework, and an open-source media and SIP stack it can run on entirely.
A real-time conversational AI framework from Agora with a visual designer, a SIP extension and turn detection, released under Apache-2.0 plus conditions that forbid competing deployments, which makes it source available rather than open source.
Added September 16, 2026. Found an error on this page? Write to reveneau@licheo.com and it gets fixed in the next weekly pass.
Running one of these inside your own environment, with your own data and your own security rules, is the kind of work Reveneau does. Read how a forward deployed engagement works.