voicebot
github.com/ruvnet/voicebot- Category
- AI Agents
- Rank
- No. 2066Tools index
- Pricing
- Open Source
- Type
- AGENT
- Builder
- ruvnet
- GitHub
- 104 stars
- Date
About
rUv's voice-driven AI bot — real-time speech-to-speech agent for hands-free interaction with LLMs.
What it does
Voicebot turns selected scenarios into outbound AI phone calls. Users choose a conversation type and voice, enter a phone number, then receive a call handled through Bland AI. Included scenarios cover interviews, coaching, support and entertainment. Prompts supply the structure and follow-up guidance.
Why it's ranked here
This is a useful, inspectable prototype for scenario-based phone conversations, with both a compact Flask application and a richer React interface. Its strongest asset is the substantial prompt library. However, several advertised agent features remain documentation-led rather than wired into the shown Python call payload.
What's good
The interaction model is easy to understand: select a scenario, voice and phone number, then start the call. Prompts are separated from request handling and contain detailed question sequences. The React version adds prompt management, settings, documentation and configurable external tools. Production instructions include Gunicorn and reverse-proxy configuration.
Tradeoffs
Phone functionality depends on a Bland AI key and external service availability. The Flask request enables call recording by default and performs only presence validation on phone numbers. Its payload sends empty tools, dynamic data, webhook and Calendly settings, despite broader intent-selection claims. Error handling can itself fail when a request raises before a response is assigned.
How to use it well
Use it as a starting point for developers prototyping outbound interview practice, coaching or scripted phone scenarios. Customize the prompt catalogue, verify recording and data policies, then harden validation and failure handling before deployment. It does not provide a complete general-purpose agent framework or a finished scheduling integration.
Technical notes+
The Python surface in app.py is a Flask form application that posts to Bland AI’s calls endpoint using BLAND_AI_API_KEY; it imports the large INTERVIEW_PROMPTS mapping from prompts.py. Dependencies are pinned in requirements.txt, including Flask, Gunicorn, FastAPI, Uvicorn, Requests, HTTPX and Pydantic. app.py may reference response before assignment after an early Requests failure. misc/calendly-main.py provides separate FastAPI examples for Calendly user, event-type and availability queries, but its root route references RedirectResponse without importing it. The React/Vite entry points are voic-tsx/src/main.tsx, voic-tsx/src/App.tsx and voic-tsx/src/pages/Index.tsx; voic-tsx/vite.config.ts configures the development server on port 8080.
Observed
- License
- MIT License
- Languages
- Python plus a JavaScript/TypeScript React application
- Installation
- Python dependencies install from requirements.txt with pip; the React application installs with npm
- Interfaces
- Browser-based web interfaces that initiate outbound phone calls through Bland AI
- Backend frameworks
- Flask for the main Python application; a separate FastAPI Calendly example is included
- Frontend stack
- React, TypeScript and Vite
- Deployment surface
- Flask development server locally, Gunicorn for production, and Vite development, build and preview scripts
Read from README.md, requirements.txt, app.py, docs/intent.md, docs/instructions.md, prompts.py, misc/calendly-main.py, voic-tsx/vite.config.ts, voic-tsx/eslint.config.js, voic-tsx/postcss.config.js, voic-tsx/tailwind.config.ts, voic-tsx/src/App.tsx, voic-tsx/src/main.tsx, voic-tsx/src/vite-env.d.ts, voic-tsx/src/pages/Index.tsx.
What it can do
Convert speech to text
User's spoken words → Text transcription
Process natural language queries with AI
Text queries or commands → AI-generated responses
Convert text responses to speech
Text from AI responses → Spoken audio output
Enable hands-free conversation with AI
Voice commands and questions → Spoken AI responses
Provide real-time voice interaction
Continuous speech input → Immediate voice responses
Tags
Tech Stack
Comments (0)
No comments yet
Editorially curated, with community endorsements as a secondary signal. Corrections welcome.