Lynx is an ambient AI agent that runs locally on a user’s device and continuously perceives the active screen, applications, and recent actions to build real‑time context. By invoking a shortcut and issuing a natural‑language command—via voice or typing—users can have the AI write, revise, summarize, reply, or organize content across apps without copy‑pasting or prompt engineering, while all data stays private on the device.
Funding
Funding not disclosed
Founders
Product
Problem
Knowledge workers spend most of their day switching between multiple applications—email, documents, chat, browsers—and must manually copy or summarize content before an AI can help. This fragmented workflow forces users to repeatedly explain context, interrupting focus and reducing productivity.
Solution
Lynx is an ambient AI agent that runs locally and continuously perceives the user’s screen, applications, and interactions. By capturing the current interface and recent actions, it builds context without requiring any copy‑paste or prompt engineering. Users invoke Lynx with a shortcut and issue a natural‑language command—by voice or typing—and the system can write, revise, summarize, reply, or organize content directly within the active workflow. Lynx can also proactively suggest next steps, plan multi‑step tasks, and execute them across apps while keeping the user in control. The entire experience stays inside the user’s existing workspace, eliminating context‑switching and allowing work to flow naturally.
Target Audience
Primary users are professionals and knowledge workers who operate across multiple desktop applications—such as managers, writers, analysts, and support staff—seeking to reduce context‑switching and accelerate routine AI‑assisted tasks.
Features
- Real‑time visual perception of active windows, tabs, and user actions to generate contextual understanding.
- Voice‑first interaction model with a single‑sentence command, removing the need for prompt formatting.
- Local processing ensures that all captured screen data stays on the user’s device for privacy and security.
- Proactive nudges that suggest unfinished tasks, replies, or organization actions based on recent activity.
- Multi‑step task planning: Lynx can break down goals, present a plan, and execute steps after user confirmation.
- Seamless execution across applications, handling files, invoking skills, and updating content without leaving the current window.
- Adjustable permission layers that let users define how much of their screen and actions the AI can access.