Skip to main content
A workflow has two faces. The editor is where you build it. The Runner is where you — or anyone you share it with — rerun it. Same workflow, same engine; different purpose.

​
The editor

The editor is the infinite canvas where you build and experiment. Drop nodes, wire them together, open a node to change its model or prompt, and run freely — a single node to test one step, or the full graph end to end. Everything is editable, all the time. Think Figma, but for AI: a place to try a model, branch a variation, and compare outputs side by side.

​
The Runner

The Runner is the published, form-driven page for rerunning a finished workflow without touching its wiring. Blend AI builds it automatically from your User Input nodes — each one becomes a field on the form. Fill in the fields, hit run, review the result. No canvas, no nodes, no editing. The same workflow is also exposed as an API, so code can trigger runs the same way.

​
Same engine underneath

A full run launched from the editor and a run launched from the Runner execute identically — same nodes, same order, same output. The only difference is purpose.
The Runner only appears once you publish. An unpublished workflow still runs perfectly well — just inside the editor.

The Runner

The published page that turns your inputs into a reusable form.

Running the full workflow

How an end-to-end run executes the whole graph.