This project provides a real-time meeting assistant demo that leverages the OpenAI Realtime API to manage a Kanban board through voice commands. It targets developers and teams looking for innovative ways to enhance meeting productivity and task management by enabling natural language interaction with a shared visual workspace. The primary benefit is the ability for multiple users to update and track tasks collaboratively using only their voice during stand-up meetings or other collaborative sessions.

How It Works

The application is built as a Go program utilizing Pion WebRTC for real-time communication and Gorilla WebSocket for signaling. It integrates Opus audio encoding/decoding and the OpenAI Realtime API. Participant audio is mixed, sent to an OpenAI Realtime peer, and function calling is employed to interpret voice commands for updating the Kanban board. This architecture allows for seamless, voice-driven manipulation of board elements, with changes broadcast live to all participants.

Quick Start & Requirements

• Install: Clone the repository, set the OPENAI_API_KEY environment variable, and run go run ..
• Prerequisites: Go 1.24 or newer, Opus library (installable via pkg-config), and an OpenAI API key.
• Access: The application is available at http://localhost:3000 by default, or a custom port can be specified using the -addr :<port> flag.
• Docs: Configuration details for initial cards, instructions, and tools are found in kanban.go.

Highlighted Details

• Enables voice-based creation, movement, tagging, updating, and deletion of Kanban tickets.
• Supports real-time updates to the shared board, with animations for card movements and confetti for completed tasks.
• The assistant is designed to ignore filler phrases, focusing only on explicit board operations.
• Customization options include initial board state, assistant instructions, exposed tools, and the Realtime model used.

Maintenance & Community

No specific details regarding maintainers, community channels (like Discord/Slack), sponsorships, or a public roadmap are provided in the README.

Licensing & Compatibility

This project is licensed under the MIT License, which is permissive and generally suitable for commercial use and integration into closed-source projects.

Limitations & Caveats

The demo explicitly states it does not include built-in authentication or access control, meaning any user accessing the application URL can join the meeting room. Additionally, background audio can be misinterpreted as board updates, requiring careful audio management (e.g., using headphones).