/voice-bot-demo

Primary LanguageJupyter Notebook

This is a quick MVP that combines high latency vocie generation from ElevenLabs with LLM reasoning capabilities and a simple vanila RAG system. I put together this demo in 24 hours to demonstrate potentail of latest AI advancements to automate call centers / voice customer support / voice sales.

Installation

Server

  • enter server directory cd voice-bot-demo/server
  • Install Python if you don't have it yet (https://www.python.org/downloads/)
  • create python environment python -m venv ./venv and activate it source ./venv/bin/activate
  • install python dependencies pip install -r requirements.txt
  • install package salesgpt pip install salesgpt (this installes other neccesary dependencies - it might be possible to just add uvicorn to requirements.txt - TBD)
  • create .env file with following environment variables:
OPENAI_API_KEY=...
ELEVEN_LABS_API_KEY=...
ELEVEN_LABS_VOICE_ID=...
  • run backend server python run_api.py -

Client

  • ensure you have node.js installed
  • enter client directory cd voice-bot-demo/client
  • npm install
  • npm run dev

Deployed

TBD

Tech stack

  • FE - Next.js
  • BE - FastAPI
  • DB - .json file stored locally in kate-sales-associate/server