Everywhere
An on-screen aware desktop AI assistant that reads the content of the application currently on screen and answers questions about it without a screenshot.
Everywhere is invoked with a keyboard shortcut over whatever is currently on screen, capturing that context automatically rather than requiring the user to take a screenshot, copy text, or switch to a separate app. Example uses in its documentation include explaining and troubleshooting an on-screen error message, summarizing a long web article, translating a highlighted foreign word, rewriting the tone of a draft email, and checking a piece of text for factual accuracy.
It connects to multiple LLM providers, including OpenAI, DeepSeek, Gemini, and Claude, and can call MCP tools to act across applications rather than only answer questions about them.
It is built on the Avalonia UI framework and targets .NET 10, and is distributed with English, Chinese, and Japanese documentation.