AI-Powered Real-Time Sales Call Assistant

An intelligent dashboard designed to help sales representatives by providing live transcriptions, sentiment analysis, and dynamic coaching tips using AI.

✨ Features

• Real-Time STT: Powered by Faster-Whisper for low-latency transcription.
• AI Coaching: Provides "Live Nudges" based on customer intent (Complaints, Inquiries, etc.).
• Local LLM: Uses TinyLlama and mistral via Ollama for privacy and offline processing.
• Post-Call Insights: Generates a structured summary and actionable auditor insights.

✨ Usage and Installation

1. Local Version (with sounddevice)

• File: app.py
• Features: Uses your computer’s microphone/audio features (sounddevice).
• Requirements: Run locally with all dependencies installed, including sounddevice.

Run locally:

git clone https://github.com/Charushi06/AI-Sales-Assistant.git
cd AI-Sales-Assistant
pip install -r requirements.txt
streamlit run app.py

Note: This version cannot be deployed on Streamlit Community Cloud because sounddevice requires local audio hardware.

2. Web Version (browser deployed)

• File: app_web.py
• Features: Compatible with Streamlit Community Cloud; works entirely in the browser.
• Requirements: Only standard Python packages from requirements.txt (except sounddevice).
• Deployed app: https://ai-sales-assistant-cg.streamlit.app/