For the first year of Aigent, every workflow lived in a form. You picked a trigger from a dropdown, added steps one at a time, and hoped the linear list in front of you matched the branching logic in your head. It worked, but it never felt like the thing it was describing.
What the canvas actually changes
The workflow canvas replaces that form with a real graph. Triggers, enrichment steps, conditions, and actions are nodes you connect directly, and the shape on screen is the shape that runs in production. When a run fails, you click the node that failed and see the exact payload it received, not a log line you have to decode.
Conditional branches were the forcing function. A linear list can fake a single if/else, but real workflows fork three or four ways depending on what a model or a webhook returns. The canvas draws every branch as its own path, so you can see at a glance which one fired for a given run.
What’s next
This is the first version. Nested sub-workflows and inline testing are next, both because they were the two things beta users asked for most.
