ConvoIntel AI

AI-powered customer conversation intelligence platform for sales and support teams.

---

Problem Statement

Organizations struggle to manually analyze large volumes of customer calls for sentiment, compliance, agent performance, and churn risks. An automated AI-driven solution is needed to generate actionable business insights efficiently.

---

Overview

ConvoIntel AI analyzes recorded customer calls using speech-to-text and large language models. It generates call summaries, tracks sentiment, evaluates agent performance, flags compliance issues, predicts churn risk, detects objections, and recommends follow-up actions — all from a single dashboard.

---

Tech Stack

Layer	Technology
Speech-to-Text	Groq Whisper
LLM Engine	Llama 3.3 70B (via Groq)
Backend	FastAPI (Python)
Frontend	React + Vite
Cloud	Render (backend) · Vercel (frontend)

---

Features

• Audio Transcription — Upload call recordings; get clean transcripts via Groq Whisper with automatic language detection
• AI Summarization — Concise call summaries generated by Llama 3.3 70B
• Sentiment Timeline — Track customer sentiment shift across the call
• Agent Scorecard — Evaluate agent performance on tone, resolution, and engagement
• Intent & Objection Detection — Identify what the customer wanted and where they pushed back
• Churn Prediction — Flag at-risk customers based on conversation signals
• Compliance Dashboard — Surface compliance violations or gaps in required disclosures
• Action Items — Auto-generate follow-up tasks from each call

---

Architecture

Audio Upload
     │
     ▼
Groq Whisper (STT)
     │
     ▼
Transcript + Language Detection
     │
     ▼
Llama 3.3 70B (LLM Analytics Engine)
  ├── Summary
  ├── Sentiment Analysis
  ├── Intent & Objection Detection
  ├── Churn Risk Score
  └── Action Items
     │
     ▼
FastAPI Backend (REST APIs)
     │
     ▼
React + Vite Dashboard (Vercel)

---

Getting Started

Prerequisites

• Python 3.10+
• Node.js 18+
• Groq API key

Backend

git clone https://github.com/your-org/convointel-ai.git
cd convointel-ai/backend

pip install -r requirements.txt

# Create .env
echo "GROQ_API_KEY=your_key_here" > .env

uvicorn main:app --reload

Frontend

cd frontend
npm install
npm run dev

---

Environment Variables

Variable	Description
GROQ_API_KEY	API key for Groq (Whisper + Llama)
BACKEND_URL	FastAPI server URL

---

Team Details

Team Information

Team Name: CallSync Domain: Artificial Intelligence & Machine Learning (AI/ML) Institution: University Visvesvaraya College of Engineering (UVCE), Bengaluru

Team Members

No.	Name	Branch
1	Akash P	Electronics and Communication Engineering (ECE)
2	Mithun S	Electronics and Communication Engineering (ECE)
3	Ashith Vinay Raj	Electronics and Communication Engineering (ECE)
4	Abhishek D Moolya	Electronics and Communication Engineering (ECE)