> ## Documentation Index
> Fetch the complete documentation index at: https://docs.tryblend.ai/llms.txt
> Use this file to discover all available pages before exploring further.

# Review and Publish

> Turn a workflow into a reusable runner page, share it, and optionally expose it through the API.

Publishing turns a canvas workflow into something other people can run. The published page uses the workflow graph you built, shows a form for the values you exposed, runs the workflow, and displays the Result outputs.

Publishing does not change how you build the workflow. It adds a reusable surface on top of it.

## What becomes public

<Steps>
  <Step title="User Input nodes become form fields">
    Add a [User Input](/workflow/nodes/input) for each value someone should fill in at run time. The input copies the field it is connected to, so the runner uses the same text, model, file, or parameter control.
  </Step>

  <Step title="User Result nodes become output fields">
    Add a [User Result](/workflow/nodes/result) for each output you want to return. Result labels become the names people see on the runner and the keys returned by the API.
  </Step>

  <Step title="Everything else stays inside the workflow">
    Models, tools, batches, collections, utility steps, and choice nodes keep running as internal workflow logic unless you expose a value with a User Input or return it through a User Result.
  </Step>
</Steps>

<Note>
  User Input nodes are the public contract. Ordinary model parameters do not automatically become form fields until you connect a User Input to them.
</Note>

## Workflow review

Before publishing, Blend AI reviews the runnable path to make sure the workflow can run as a published workflow.

The review checks that:

* The workflow has at least one User Result node.
* The runnable path can execute without graph errors.
* The runnable path has a source: a User Input, Batch, or Collection node.
* Every User Input node on the runnable path is connected to a real downstream field.
* User Input labels normalize to unique API keys.
* User Result labels normalize to unique output keys.
* The runnable path uses supported workflow node types.

If something fails review, fix the highlighted node or label on the canvas, then publish again.

## Publish settings

The publish dialog is split into four sections:

| Section          | What you configure                                                       |
| ---------------- | ------------------------------------------------------------------------ |
| Listing          | Workflow title, description, cover image, categories, and tags.          |
| Access & pricing | Visibility, shared recipients, and the creator fee for public workflows. |
| Examples         | Completed sample runs that show what good inputs and outputs look like.  |
| Details          | API slug and release notes for the published version.                    |

After a successful publish, the dialog also shows an **API** tab with generated `curl`, JavaScript, and Python examples for that workflow.

## Visibility

| Visibility  | Who can run it                                                     |
| ----------- | ------------------------------------------------------------------ |
| Private     | Only you.                                                          |
| Shared      | Specific people you invite by email.                               |
| Public      | Anyone with access to the public workflow page or library listing. |
| Open source | Public, with the graph shared so others can copy and build on it.  |

Public and open-source workflows require a creator username so the listing can show who made the workflow.

## Pricing

Public workflows can include a **creator fee**. The runner pays for the workflow's compute plus the fee you set.

You can set the fee directly as a dollar amount or as a percentage of the estimated workflow cost. Blend AI converts it to credits for the buyer.

Private and shared workflows are not marketplace listings, so they do not charge a public creator fee.

## Republish and versions

Publishing creates an immutable version snapshot of the graph, pricing, and selected examples. If you change the workflow, publish again to make the latest behavior live.

Existing published workflow settings can be reused, and the current version pointer moves to the newly published version.

## API access

The API is part of publishing, but it is not the only reason to publish. Use it when you want the same published workflow to run from code.

<CardGroup cols={2}>
  <Card title="User Input" icon="arrow-right-to-bracket" href="/workflow/nodes/input">
    Expose values as runner form fields and API input parameters.
  </Card>

  <Card title="User Result" icon="flag-checkered" href="/workflow/nodes/result">
    Return workflow outputs to the runner and API.
  </Card>

  <Card title="API Reference" icon="code" href="/api-reference/overview">
    Start, stream, and poll published workflow runs from code.
  </Card>
</CardGroup>