Skip to content

Nightkilller/MeetMind

Repository files navigation

MeetMind — AI Meeting Intelligence Platform

Microsoft Build AI Hackathon 2026 | Theme: AI at Work: Productivity & Teamwork Reimagined

Problem Statement

Professionals spend 31 hours per month in unproductive meetings. 70% of action items discussed in meetings are never followed up on. Existing tools either transcribe without understanding, or summarize without extracting next steps. MeetMind solves all three gaps in one platform.

Solution Overview

MeetMind is a real-time AI meeting copilot that:

  • 🎙️ Records and transcribes meetings using Groq Whisper Large V3
  • 🧠 Analyzes transcripts using Groq Llama 3.3 70B to extract structured insights
  • ✅ Automatically generates action items with owners, priorities, and due dates
  • 📧 Drafts follow-up emails in one click
  • 🔍 Provides a searchable dashboard of all past meetings

Live Demo

🔗 Live App: https://meet-mind-bay.vercel.app 🎥 Demo Video: https://youtu.be/fflYYIb3ov0 📁 GitHub: https://github.com/Nightkilller/MeetMind

Test Credentials

The app uses Google OAuth via Clerk.

Field Value
Email ranu92352@gmail.com
Password Meetmind123@

How to login:

  1. Go to https://meet-mind-bay.vercel.app
  2. Click Sign In to Get Started
  3. Click Sign in with Google
  4. Enter the email and password above
  5. Dashboard loads with 3 demo meetings pre-loaded

Note: 3 demo meetings are pre-loaded so you can immediately see all features — transcription, summaries, action items, health scores, and email drafts — without recording anything.

Architecture

Browser (Next.js 16 App Router)
  → Groq Whisper Large V3   → Real-time Transcription
  → Groq Llama 3.3 70B      → Summary, Action Items, Email Draft
  → MongoDB Atlas           → Persistent Storage
  → Clerk v7                → Google OAuth
  → Vercel                  → Deployment

Microsoft AI Stack Used

Service Usage
Azure AI Inference — GitHub Models (GPT-4o) Sentiment analysis for Meeting Health Score via models.inference.ai.azure.com
GitHub Copilot AI-assisted development, component scaffolding, debugging

AI Tools Used in Development

Tool Purpose
Azure AI Inference (GitHub Models) GPT-4o sentiment analysis
GitHub Copilot Inline code suggestions and autocompletion
Groq Whisper Large V3 Real-time audio transcription
Groq Llama 3.3 70B Meeting analysis, action items, title generation, email draft
Claude (Anthropic) Architecture planning and prompt engineering
Cursor AI-assisted full-stack code generation

Tech Stack

Layer Technology
Frontend Next.js 16 (App Router) + TypeScript
Styling Tailwind CSS v4 + Custom CSS (glassmorphism)
Transcription Groq Whisper Large V3
AI Analysis Groq Llama 3.3 70B
Database MongoDB Atlas (Mongoose)
Auth Clerk (v7)
Animations Framer Motion
Data Fetching SWR
Deployment Vercel

Setup Instructions

Prerequisites

  • Node.js 18+
  • Groq account — free at console.groq.com
  • MongoDB Atlas account (free tier)
  • Clerk account — free at clerk.com

Local Development

git clone https://github.com/Nightkilller/MeetMind.git
cd meetmind
npm install --legacy-peer-deps
cp .env.example .env.local
# Fill in your Groq API key and MongoDB credentials
npm run dev

Open http://localhost:3000

Environment Variables

See .env.example for required variables. You will need:

  • GROQ_API_KEY (Free from console.groq.com)
  • MONGODB_URI (Free Atlas cluster)
  • NEXT_PUBLIC_CLERK_PUBLISHABLE_KEY & CLERK_SECRET_KEY (Free from clerk.com)
  • GITHUB_AI_TOKEN (Free from github.com/marketplace/models — no credit card needed)

Features

  • Context-Aware AI Templates: Select meeting type (Standup, Planning, Client Call) to dynamically bias Groq's analysis prompt.
  • AI Meeting Title Generator: Automatically generates a concise 8-word title from the transcript using Groq if left blank.
  • Meeting Health Score: A dynamic grading algorithm out of 100 based on positive/negative sentiment, action items, key decisions, and duration.
  • Smart Action Item Reminders: The dashboard proactively alerts you with a red banner when action items are due today or overdue.
  • Live Transcript Animation: A stunning "wow-factor" word-by-word simulated stream mimicking real-time typing.
  • EmailJS Integration: Send beautifully formatted follow-up emails directly to multiple comma-separated recipients without leaving the app.
  • Live Outlook Calendar Sync: Sync meeting summaries, decisions, and action items directly to Office 365 or Outlook.com calendars, or download .ics files.
  • ✅ Real-time audio recording in browser (MediaRecorder API)
  • ✅ Groq Whisper Large V3 transcription (fast, accurate)
  • ✅ Groq Llama 3.3 70B meeting analysis (summary, decisions, action items)
  • ✅ Action items with owner, priority, and due date
  • ✅ Searchable meeting history dashboard
  • ✅ Secure authentication (Clerk)
  • ✅ Responsive design (mobile + desktop)
  • ✅ Microsoft Azure enterprise UI (Light Theme)

Project Structure

meetmind/
├── app/
│   ├── layout.tsx                    # Root layout
│   ├── page.tsx                      # Landing page
│   ├── dashboard/page.tsx            # Dashboard
│   ├── meeting/new/page.tsx          # New meeting recorder
│   ├── meeting/[id]/page.tsx         # Meeting detail (3-panel)
│   └── api/                          # API routes
│       ├── meetings/                 # CRUD
│       ├── transcribe/               # Groq Whisper
│       ├── analyze/                  # Groq Llama 3.3 70B
│       └── generate-title/           # AI Title Generation
├── components/
│   ├── ui/                           # Button, Card, Badge, Modal, Spinner, Toast
│   ├── layout/                       # Navbar, Sidebar, PageWrapper
│   ├── meeting/                      # MeetingCard, MeetingRecorder, TranscriptViewer,
│   │                                 #   SummaryPanel, ActionItemsList, EmailDraftModal, SpeakerTag
│   └── dashboard/                    # StatsBar, RecentMeetings, SearchBar
├── lib/                              # mongodb, groq, prompts, utils
├── models/                           # Meeting, ActionItem, User (Mongoose)
├── hooks/                            # useMeetingRecorder, useMeetings, useTranscript
└── types/                            # TypeScript interfaces

All AI-generated code was reviewed, modified, and validated by the team. Prompts, architecture decisions, UI design, and business logic were created with meaningful human judgment throughout.

Team

Team Name: BerryBytes

Name Role
Aditya Gupta Leader / Full Stack Developer
Dhairya Bhatnagar Team Member / AI Integration

License

MIT — see LICENSE

About

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

 
 
 

Contributors

Languages