bolna-ai/bolna
Conversational voice AI agents
What it solves
Bolna solves the complexity of building production-ready, voice-driven conversational assistants. It handles the difficult task of orchestrating multiple specialized services—such as speech recognition, large language models, and text-to-speech—into a single, coherent conversation flow.
How it works
The platform acts as an orchestration layer that connects different providers via websockets. It typically follows a pipeline: transcribing audio input (ASR), processing the text through an LLM, and synthesizing the response back into audio (TTS). It supports telephony integration through providers like Twilio or Plivo to enable real-world phone call interactions.
Who it’s for
It is designed for developers looking to build voice-first AI applications, ranging from simple text-only assistants to complex, telephony-integrated voice agents.
Highlights
- Multi-provider support: Integrates with ASR (Deepgram, Azure), LLMs (OpenAI, Llama, Mistral), and TTS (ElevenLabs, AWS Polly, Cartesia).
- Telephony integration: Supports initiating and managing calls via Twilio and Plivo.
- Flexible orchestration: Can be used as a programmatic Python library or via a hosted API.
- End-to-end pipeline: Manages the full lifecycle from audio input to synthesized audio output.
- Extensible architecture: Allows developers to add custom telephony and provider handlers.