Teach once. Generate safely.
Turn one verified robot demonstration into reusable, editable automation.
1 · Identify — Robot status
Control interface
Unknown — upload a Kortex JSON in step 2 to confirm
A successfully parsed Kortex sequence export is direct evidence the arm speaks Kortex. Nothing else about the robot is assumed from it.
Brand & model
Needs confirmationThe Kortex export contains no model information, and a photo alone cannot reliably identify the exact model — so this stays "Needs confirmation" until you confirm it. For AI photo identification, use the Try it flow.
2 · Learn — Workspace from demonstration
Upload a sequence export that ran successfully on the real arm. The file is parsed in your browser and never modified.
3 · Create — New pick-and-place task
Confirm every Workspace Profile value above before creating a task — only confirmed parameters are used for generation.
A tabletop photo cannot provide reliable robot X/Y coordinates (that needs camera calibration, AprilTag/ArUco markers, or a depth camera). Enter coordinates in the robot's base frame — jog the arm in Kortex to read them off, or reuse values from the demonstration table above.
4 · Validate — Timeline, path & safety
The timeline and path preview appear once the workspace profile is confirmed and the task inputs are complete.
5 · Export