an intelligent meeting assistant that transcribes, analyzes, and extracts action items from meetings. powered by whisper, llama3.2, and a modern next.js stack.

• meeting recording and transcription
• ai-powered meeting analysis
• automated task extraction
• meeting insights and summaries
• task management integration
• dark/light mode support

• next.js 14 (app router)
• typescript
• tailwind css
• shadcn/ui components
• clerk authentication
• local storage for data
• openai whisper for transcription
• ollama (llama 3.2) for local ai analysis

next.js
react-based framework with support for app router, server components, and api routes — used for fast and scalable frontend + backend logic.

clerk auth
handles user authentication with sessions, mfa, and identity provider support — integrated for secure, production-ready auth.

whisper
used for both live and post-meeting transcription. supports multi-language input, speaker separation, and solid accuracy.

ollama + llama 3.2
runs locally using ollama, powering offline and fast analysis. extracts meeting summaries, insights, and todo tasks from the transcript.

to run llama3.2 locally:

ollama run llama3.2

1. clone the repo

git clone https://github.com/yourusername/iris.git
cd iris

2. install dependencies

npm install

3. setup environment variables

create a .env.local file and add:

OPENAI_API_KEY=your_openai_api_key
NEXT_PUBLIC_CLERK_PUBLISHABLE_KEY=your_clerk_publishable_key
CLERK_SECRET_KEY=your_clerk_secret_key

4. start dev server

npm run dev

• OPENAI_API_KEY
• NEXT_PUBLIC_CLERK_PUBLISHABLE_KEY
• CLERK_SECRET_KEY

pull requests are welcome. feel free to fork, tweak, and build.