Glass is a desktop application that extends your digital mind by capturing and understanding your screen activity and audio in real time.
The tool addresses the problem of losing context and action items during work by running locally on your computer to observe what you see and hear, then using AI to surface summaries, answers, and structured knowledge from that context. It works by capturing screen and audio data, processing it through configurable language models, and making that information queryable through a chat interface. The application supports multiple AI backends including OpenAI, Gemini, and local models via Ollama and Whisper.
Glass suits developers and knowledge workers who want to augment their memory and productivity without sacrificing privacy. It is designed for local-first operation, meaning data stays on your machine rather than being sent to external servers. The tool is particularly useful during meetings where it can generate real-time notes and summaries. Users can bring their own API keys for supported language models or use the project's free tier. The application is built with Electron and Next.js, making it a desktop-native experience rather than a web application.
The project is undergoing a full code refactor and modularization effort, with the maintainers prioritizing architectural improvements before addressing major open issues. Contributions are actively welcomed through the standard GitHub workflow, with a dedicated contributing guide available for those interested in participating.
