Skip to main content
There are two ways to put an agent in front of other people: share a read-only link to the workflow itself, or deploy the agent as an app that others run from a simple form. Use a read-only link to show how a workflow is built; use an app to let people run it without touching the graph. A share link opens your agent on a locked canvas. Viewers can pan around and inspect the nodes and the last run’s results, but they can’t edit anything.
1

Turn on the public link

In the studio, open the share controls and enable the public link. Orisu generates a unique URL you can copy and send.
2

Send the link

Anyone with the link can open the read-only view. They see the agent’s name, a summary of the nodes in the workflow, and the results from its last run.
3

Let viewers clone it

From a shared link, a signed-in viewer can clone the workflow into their own workspace and tweak it. Cloning copies the graph so they start from your working version instead of from scratch.
Sensitive config fields (credentials, API keys, and private URLs) are stripped from the public view. Only the workflow structure and its results are shown.
A share link points at your agent as it is now, not at a snapshot of it. Keep editing in the studio and viewers see your latest saved changes on the same URL, including the results of any run you do afterwards. You never need to re-share a link to update what people see. (Deployed apps work the other way around — see below.) The link is stable, so it’s safe to send once and keep working. Turning sharing on for an agent that’s already shared gives you back the same URL rather than a new one. If you do need to invalidate a link — you sent it to the wrong person, or the workflow has moved on — regenerate it. The old URL stops working immediately and a new one takes its place, so anyone still using the old link loses access. You can also switch the public link off entirely.

​
Deploy as an app

Deploying turns your agent into an app: a form-based interface where people fill in inputs and get results, without seeing or editing the node graph. This is the way to hand a workflow to teammates who don’t need to understand how it’s built.
1

Choose inputs

Pick which input nodes (text prompts and uploads) become fields in the app’s form. Each becomes a labeled field your app users fill in. Other input and brand nodes stay as context that drives the agent but isn’t shown to users.
2

Choose outputs

Pick which result nodes are shown to app users. Nodes that generate images, video, text, audio, or file exports can be surfaced as outputs.
3

Name it and set visibility

Give the app a name, description, and cover image, then choose who can access it:
4

Deploy

Publish the app. App users open it, fill in the form, and run it. They see live progress and the outputs you chose, never the underlying graph.
Running an app uses the same execution as the studio, including loop iterations and human-review pauses. App users can re-run with new inputs and revisit earlier runs.
Deploying pins the app to a saved version of your agent. If you keep editing afterward, deploy again to push your latest changes to the app.