🤖 Takumi

Takumi is an open-source, AI-powered meeting assistant platform built with the modern web stack. It bridges the gap between real-time video conferencing and intelligent post-meeting analysis.

With Takumi, users can create custom AI Agents with specific personas/instructions, invite them to meetings, and interact with them afterward to query transcripts, summaries, and action items.

🚀 Features

• 📹 Real-time Video Meetings: High-quality video calls powered by Stream Video.
• 🤖 Custom AI Agents: Create agents with unique personalities and instructions (e.g., "Note Taker", "Technical Reviewer").
• 📝 Automated Transcription & Summarization: Meetings are automatically transcribed and summarized using OpenAI and Inngest.
• 💬 Interactive AI Chat: Chat with your AI agents after the meeting. They have full context of the transcript and can answer specific questions about what happened.
• 🔐 Secure Authentication: Built-in auth using Better Auth.
• 💸 Subscription Ready: Integrated with Polar for managing premium features and subscriptions.
• ⚡ Reactive UI: Built with Shadcn UI, Tailwind CSS, and React 19.

🛠️ Tech Stack

• Framework: Next.js 15 (App Router)
• Language: TypeScript
• Database: PostgreSQL (via Neon)
• ORM: Drizzle ORM
• Video & Chat: Stream
• AI & LLMs: OpenAI
• Background Jobs: Inngest
• Authentication: Better Auth
• Payments: Polar
• API Layer: tRPC

---

⚡ Getting Started

Follow these steps to get Takumi running locally on your machine.

Prerequisites

Ensure you have the following installed:

• Node.js (v18 or higher)
• npm or pnpm
• PostgreSQL Database (We recommend Neon Serverless)

1. Clone the Repository

git clone [https://github.com/vasanthrs-dev/takumi.git](https://github.com/vasanthrs-dev/takumi.git)
cd takumi

2. Install Dependencies

npm install

3. Configure Environment Variables

Create a .env.local file in the root directory. You will need credentials from the services listed below:

# 1. Database (Neon/Postgres)
# Get this from your Neon Dashboard
DATABASE_URL="postgresql://user:password@host/dbname?sslmode=require"

# 2. Authentication (Better Auth)
# The URL where your app is running (http://localhost:3000 for dev)
BETTER_AUTH_URL="http://localhost:3000"
# Generate a random string: openssl rand -base64 32
BETTER_AUTH_SECRET="your_generated_secret_here"

# 3. OAuth Providers (Required for Sign In)
# You need to set these up in Google Cloud Console or GitHub Developer Settings
GITHUB_CLIENT_ID="your_github_client_id"
GITHUB_CLIENT_SECRET="your_github_client_secret"
# (Optional: Add Google if you are using it)
GOOGLE_CLIENT_ID="your_google_client_id"
GOOGLE_CLIENT_SECRET="your_google_client_secret"

# 4. Stream (Video & Chat)
# Get these from the GetStream.io Dashboard
NEXT_PUBLIC_STREAM_VIDEO_API_KEY="your_stream_api_key"
STREAM_VIDEO_SECRET_KEY="your_stream_secret_key"

# 5. AI Services (OpenAI)
# Get this from platform.openai.com
OPENAI_API_KEY="sk-..."

# 6. Payments (Polar)
# Get this from polar.sh settings
POLAR_ACCESS_TOKEN="your_polar_access_token"

# 7. Background Jobs (Inngest)
# Optional for local development (uses Inngest Dev Server default)
# Required for production
INNGEST_EVENT_KEY="local"
INNGEST_SIGNING_KEY="local"

4. Database Setup

Push the database schema to your PostgreSQL instance using Drizzle.

npm run db:push

5. Run the Application

You need to run the development server and the Inngest dev server (for handling background workflows like summarization).

Terminal 1: Next.js App

npm run dev

Terminal 2: Inngest Dev Server

npx inngest-cli@latest dev

Open http://localhost:3000 to see the app. Open http://localhost:8288 to see the Inngest dashboard.

---

🔄 Architecture Overview

The Meeting Lifecycle

1. Creation: A user creates a meeting and assigns an Agent to it.
2. Live Session: The user joins the video call powered by Stream. The session is recorded.
3. Processing: Once the meeting ends, a webhook triggers an Inngest function.

• The system fetches the transcript.
• It identifies speakers (Users vs. Agents).
• It uses OpenAI to generate a structured summary.

4. Interaction: The user can visit the meeting detail page and chat. The chat messages are processed by an Inngest function that feeds the meeting summary + agent instructions to OpenAI to generate a response.

---

📂 Project Structure

src/
├── app/              # Next.js App Router pages (auth, dashboard, call)
├── components/       # Reusable UI components (Shadcn)
├── db/               # Drizzle schema and connection logic
├── inngest/          # Background job definitions (AI processing)
├── lib/              # Utility libraries (Stream, OpenAI, Utils)
├── modules/          # Feature-based architecture
│   ├── agents/       # Agent management logic
│   ├── auth/         # Authentication views
│   ├── call/         # Video call UI & logic
│   ├── dashboard/    # Dashboard layout & components
│   ├── home/         # Landing page & public views
│   ├── meetings/     # Meeting management & transcripts
│   └── premium/      # Polar subscription logic
└── trpc/             # tRPC API definition

---

🤝 Contributing

We love contributions! Takumi is open source and built for the community.

1. Fork the repository.
2. Create a Feature Branch (git checkout -b feature/AmazingFeature).
3. Commit your changes (git commit -m 'Add some AmazingFeature').
4. Push to the branch (git push origin feature/AmazingFeature).
5. Open a Pull Request.

Development Scripts

• npm run dev: Start the dev server.
• npm run db:studio: Open Drizzle Studio to view your database data visually.
• npm run dev:webhook: Start an ngrok tunnel (useful for testing Stream webhooks locally).

---

📄 License

This project is licensed under the MIT License. Feel free to use, modify, and distribute it as you wish.