Agent activity usually lives in chat windows, logs and endless “working…” indicators. This project takes a different route: a desktop app that turns an AI team’s work into a living subway map.
Each agent becomes a coloured circle with initials. Routes start together at a central dispatch station, then branch out as work happens. Tool calls and observed visits to servers or information sources become stops. Shared destinations become interchanges.
The movement follows actual events from a local agent system. Tracks appear as journeys unfold, so the map shows recorded activity rather than a decorative progress animation.
A few highlights:
- A network with a memory. Completed journeys remain as faded lines, gradually building a map of past work.
- Clickable trips. Select an old route to inspect its timeline, individual stops and final result.
- Signals that mean something. A red railway signal flags approvals and clarification questions, with answers handled directly inside the app.
- An onboard conversation. Chat with the lead agent to create specialists and dispatch tasks without switching apps.
Visually, the inspiration is vintage transit design: warm off-white paper colours, muted inks, rounded routes, shared station markers and angled labels. There’s a dark version too, for the night service.
It’s a working concept built around a simple idea: watching agents work should feel less like reading a server log and more like following a journey.
Next stop: task complete.