Use input and output nodes

Input nodes bring assets into a workflow; output nodes define where generated files land so you can download them or send them to connected systems. Together, they bookend almost every production workflow in Workflow Builder.

How input and output nodes work

Input nodes expose one or more assets to the rest of the workflow. You can add these assets from your local drive (upload one file or many) or from cloud-connected locations your organization has configured. Supported cloud sources commonly include Adobe Experience Manager (AEM) Assets and other Adobe cloud storage integrations, which let you pull content from services such as Adobe Cloud Storage and Frame.io when available in your environment.

Output nodes receive the files your workflow produces. They’re the handoff point for exporting finished assets back to local download paths and supported cloud systems. This allows workflow runs to pull and send assets between the Workflow Builder and your Digital Asset Management system or review tools.

When you upload assets from local files, the workflow processes them in the cloud; finished files appear on output nodes according to how you initially connected the workflow. You can download finished products from an output node’s location to local storage. Any file(s) generate by the workflow are also listed in your generation history- Open Canvas View and use the right-hand toolbar and then open history. From here, you can review, inspect, and download outputs without hunting for them only at individual nodes.

When to use which

• Start a branch with inputs when you need images, video, text, or documents from disk or cloud to feed downstream nodes.
• End a branch with outputs when you need predictable delivery of finals—whether for local archival or push to cloud storage your team uses for approval and distribution.
• Use generation history when you want a run-level view of everything the workflow produced, especially for comparison across runs or quick downloads.

Working with local uploads

1. Add the appropriate Input node (for example, Input images, Input videos, or Input text) to the canvas.
2. In the node, upload one or multiple files from your device. Multi-asset inputs are supported where the node catalog allows batch or list inputs.
3. Connect the node’s outputs to processing nodes (such as adjustment, generative, merge, and so on).
4. Add an Output node (or the output type your workflow requires) and connect the final results into it.
5. Run the workflow. Processed files are written to the output node’s destination; use the node or generation history to download to local storage.

Working with cloud sources and destinations

1. Ensure your account and organization have access to the relevant integrations (AEM, Adobe cloud storage, Frame.io, or others surfaced in the product).
2. On an input node, choose the cloud source instead of local upload, then pick the asset or folder the workflow should use.
3. On an output node, select the cloud destination so merged or generated files are imported/exported through Workflow Builder into those systems.
4. Run or publish the workflow as usual. Permissions and entitlements still apply; if a source does not appear, verify connectivity and catalog availability for your workspace.
5. Exact control names and supported storage types can vary by node and release. If the UI differs slightly from this guide, follow the labels shown in Workflow Builder for your build.

Providing inputs programmatically (API)

• When you invoke a published workflow through the API instead of running it from the canvas, inputs are supplied in the request payload rather than through upload dialogs. The same logical inputs—binary assets and simple values—map to parameters your workflow exposes when published.
• Binary inputs (images, video, documents, and similar file-backed assets) are typically passed as pre-signed URLs that point to objects your integration already stored in a supported location. The service fetches those assets at run time using the URLs you provide.
• Simple types (for example, plain text, flags, or other scalar fields exposed by input or configuration nodes) are sent inline in the payload as structured fields—strings, numbers, or booleans—alongside any binary references.
• Parameter names, nesting, and required fields align with how the workflow was authored and published; your integration must supply values for every required input the published workflow declares.
• For authentication, request and response formats, batch execution, error handling, and end-to-end examples of integrating and scaling workflows via API, see the Workflow Builder API on Adobe Developer.

Finding outputs after a run