AI coding agents are changing how software is built, but they still struggle with one fundamental problem: understanding what users are referring to on screen. Developers waste time writing long prompts, taking screenshots, and correcting AI instead of simply pointing. We believe this communication bottleneck will become more important than model intelligence.

Supamaus is a gesture-based context layer that lets users point, circle, and gesture naturally while interacting with AI agents like Claude Code and Codex. Through MCP, it gives any compatible agent the visual context needed to understand user intent.