Working with flows

Enterprise Single-tenant

Understanding the flow editor interface can help you efficiently create flows.

The default interface on the Visual Editor tab includes these elements:

  • The canvas, where you create the diagram of your flow by adding steps. Connections between steps show how records move through the pipeline. Use the canvas controls to pan and zoom, select steps, and open each step’s configuration.

  • The Input samples panel, where you can add sample files to test run your flow.

  • The Modules panel, where you import and manage modules. See Using flow modules for details.

  • The Events panel, where you can add pre-flow or post-flow events for upstream and downstream integrations. See Integrating flows for details.

  • The Add steps panel lists all available flow steps, categorized under Map, Filter, or Reduce. See Flow steps for usage and configuration details for each step.

In addition to the visual editor, use the following tabs to understand or modify your flow:

  • JSON viewer — Shows the flow definition as JSON so you can inspect or edit the underlying structure directly. Use it when you need precision editing, bulk changes, or to compare definitions across environments.

  • Flow settings — Configure input, output, and runtime settings.

  • Summary — View recent flow runs and their logs.

Creating flows

Use the canvas in the visual editor to build your flow.

See Planning flows for advice on how to think about and order steps.
  1. In the header, click the initials icon and select Switch to advanced view.

  2. In the left sidebar, click the All apps icon, then select Flow. The flow editor opens.

  3. Click Create New, and define a name for your flow

  4. Define where in the file system to store the flow and its associated files.

    1. In the Where field, click the folder icon.

    2. From the second dropdown, select a workspace.

    3. In the file explorer, select a connected drive, or use the Instabase Drive.

    4. Select an existing folder, or click New folder. If adding a folder, name it then click Create folder.

    5. Click Open folder.

  5. Click Create.

  6. Add steps to build your flow. Click + Add flow step to get started, or drag and drop steps onto the canvas from the Add steps panel.

  7. As you add steps, their configuration panels open. Configure the step and, if required, link a module. To close out the configuration panel, click the X icon, or click on the canvas.

    You can rename steps to make flows easier to understand, such as naming filter steps according to what’s being let through. In the step configuration panel, click the edit (pencil) icon next to the step name.

Testing flows

You can test your flow against sample documents to assess how its performing.

Before you begin

Gather and upload files that are representative of the types of documents the flow is built to process. If using classification, ensure you have documents of each class.

  1. In the header, click the initials icon and select Switch to advanced view.

  2. In the left sidebar, click the Explorer icon. The file explorer opens.

  3. In the workspace list, select the target workspace.

  4. In the drive list, select the Data Drive or another connected drive.

    The Data Drive is part of your 1 TB of included storage and intended for use as a file input source.
  5. Select a folder or click New > New folder to create one. Using dedicated, clearly labeled folders helps organize your testing documents.

  6. Select or drag and drop files or folders into the upload dialog, then click Start upload.

  1. In the Input samples panel, click the + icon.

  2. In the file explorer, select the folder containing your sample files. Click Open folder.

  3. On the Flow settings tab, use the Input Folder dropdown to select your input sample folder.

  4. Configure other flow settings as needed.

  5. Click Run flow.

You can monitor the flow run progress and logs from the Summary tab.

Flow settings

On the Flow settings tab, choose input and output locations, processing mode, and runtime options. These settings apply when you run the flow.

SettingDescription
Input FolderRequired. Select the folder that contains the files to process for this run.
Compare Against Ground Truth SetWhen enabled, compares the flow output to a ground truth dataset so you can check accuracy.
Output FolderThe folder where the flow writes results. Use the browse control to choose a path.
Do Not Overwrite OutputWhen enabled, avoids overwriting existing files in the output folder when a result file would have the same name.
Processing ModeSelect how the flow runs (for example, Agent Mode).
Runtime ConfigOptional JSON object for advanced runtime parameters. The editor starts with an empty object ({}) that you can extend.