Winds: climatology
Flood engine: loading
Share this page with teammates: everyone who opens it joins the same room, sees each other's cursors and hands, regions, answers and simulations.
Each gesture gets a few seconds of samples from your webcam. A small neural network trains on your hand landmarks right here, then runs live alongside the built-in gestures. Record Neutral too, so ordinary hand poses don't fire actions.
| Open palm, move | Spin the globe |
| Pinch and drag | Draw a region (Ask) · draw a storm track or grab the globe (What-If) |
| Pinch on a button | Press it |
| Two hands apart / together | Zoom in / out |
| Fist, move up / down | Raise / lower the sea (What-If) |
| Peace sign, hold | Switch mode |
| Drag | Spin, or draw with the active tool |
| Scroll / pinch | Zoom |
| Shift + drag | Draw a region from any tool |
| “Go to Tokyo” | Fly to a city |
| “What if the sea rose 3 metres” | Jump into the simulation |
| “Ask mode” / “What-if mode” | Switch mode |
| “Show me this region in 2000 vs 2026” | Satellite swipe compare |
| Any question | The agent plans, calls data tools and draws layers |
Hand tracking and voice need camera and microphone access, which the local version provides (run it on your computer; see README). Live winds, earthquakes, NASA imagery, World Bank data, named rooms and Google 3D Tiles also come through the local server. In a headset, use the VR / AR button: grab the globe to spin it, pinch with both hands to scale, point and pinch to draw.