Build a Wingz Workflow Manually to Automate Tasks

Overview

You can build a workflow in four steps: create it, configure what triggers it, add one or more Agent nodes to perform tasks, and define what happens once it ends. These four steps apply to every workflow. Beyond them, you can optionally add Logic nodes to branch, route, or run agent tasks in parallel. Each step corresponds to one of the building blocks of Wingz workflows. To build a complete workflow, you must configure each of these on the Workflows canvas.

You can build a workflow to run its agent tasks either sequentially or parallely. A sequential workflow runs one Agent node after another, where each task waits for the previous one to complete. A parallel workflow, in contrast, runs more than one Agent task at the same time, then brings the results back together before the workflow continues.

For example, a sequential workflow could summarize a paused A/B test and email the result, one step naturally following the next. A parallel workflow could analyze a heatmap and review session recordings for the same campaign at the same time, then combine both findings into a single report. This allows independent tasks to run without waiting for one another. 

Build a Sequential Workflow

Step 1: Create a Workflow

To create a new workflow from the Workflows dashboard:

1. Log in to your Wingify account.
2. From the left navigation panel, go to Wingz > Workflows. This opens the Workflows dashboard, which displays a workflow template gallery and your existing workflows list (if there are any).
3. From the dashboard, choose a starting point for the workflow:

   1. Use a template:
      Click a listed template, for example, Automated Campaign Debrief or Conversion Drop Alert, to open the workflow canvas pre-filled with a Start node, Agent node, and End node already connected for that use case. Choose this when your automation is close to a common pattern. You can then edit the pre-filled nodes instead of building the flow from scratch.
      
      Click See All to browse the full set of available templates, or use the Search and Add filter options on the dashboard to find one that matches what you need.

      Tip: Starting from a template that closely matches your requirement is usually faster than building a workflow from scratch. You can rename, reconfigure, or extend any template's nodes once it opens on the canvas.
   2. Start from scratch:
      Use this option or click Create, to open a blank canvas with a Start node and an End node already placed on it. Choose this when your automation doesn't match an existing template, or when you want to configure every node yourself from the start.

Once the Workflow canvas opens, the Wingz Workflow Assistant panel sits on the left, the node palette sits next to it, and a Workflow settings panel opens on the right, where you can edit the workflow's Name and add a Description.

Step 2: Define the Start Trigger

The Start node determines when the workflow runs or initiates. Click the Start node on the canvas to select it, then choose a Trigger Type in the right panel. You can choose from two available options:

• Select Time Based to run the workflow on a recurring schedule, then set Schedule to Daily, Weekly, or Monthly.
• Select Condition Based to run the workflow when a campaign matches conditions you set, then configure the following:
  • Campaign Type
  • Campaign Event
  • Campaigns

For details on how each trigger type works and what its fields control, see the Start Node section in Components of a Workflow.

Step 3: Add Agent Nodes

An Agent node runs a Wingz agent to perform one task in the workflow, such as analyzing a heatmap or generating a PDF. To add an agent node:

1. Drag an Agent node from the Wingz section of the node palette and drop it onto the canvas.

   Note: There is no limit to the number of Agent nodes you can add to a workflow.
2. Connect it to the Start node, or to the output of another node already on the canvas. To delete the connection between two nodes, click on the connecting wire and press the Delete key on the keyboard.
3. While the Agent node is selected, from the node settings panel on the right, choose whether to use an existing agent or create a new one.
4. To use an existing agent:
   1. Select an agent from the Select an agent dropdown.
   2. Optionally, add context in the Additional Instruction field to guide this specific run without changing the agent's own configuration. For example, if the agent's goal is to analyze campaign performance, adding focus on the mobile segment only narrows this one run. For more information, see Agent Node.
   3. Review the Agent Details shown below the dropdown. Click Edit Agent to update:
      1. Name
      2. Goal
      3. Prompt
         
         Based on the Goal and Prompt you provide, Wingz automatically determines and uses the appropriate tools to execute the task.
5. To create a new agent:
   1. Click Create New Agent to define a new agent directly within the node.
   2. Configure the new agent's Name, Goal, and Prompt.
   3. Click Create.
   4. Review and edit the agent’s details.
6. Click Save.

Step 4: Configure the Workflow's End Action

The End node determines what happens when the workflow ends. To configure the end action:

1. Click the End node on the canvas to select it.
2. In the right panel, set Output Type to:
   1. Do Nothing to end the workflow without any further action, or
   2. Notify to send an email when the workflow ends
3. If you select Notify, select the Email checkbox under Notify me on, then add recipient email addresses. Add each address on its own line, or separate multiple addresses with commas. Each recipient appears as a removable chip above the field.
4. Select Generate AI email summary to have Wingz summarize the workflow's results, including PDF reports, in the notification email. Leave it unchecked to send the raw output instead.
5. If you selected Generate AI email summary, optionally edit the AI summary instructions field to change what the summary covers. You can also change how it is structured, or click Reset to default to restore the original template.

For example, set Output Type to Notify with your own email address to get a summary every time the workflow completes, or leave it at Do Nothing for a workflow that only needs to update data in the background without alerting anyone.
 

Step 5: Save, Test, and Publish the Workflow

Once all nodes are configured, save your changes, verify that the workflow behaves as expected, and then activate it. Complete the following steps:

1. Click Save to save your changes without activating the workflow.

   Note: Before clicking Save, confirm every node you've added is connected to the workflow, from the Start node through to the End node, with no isolated or dangling nodes left on the canvas.

2. Click Test on the top-right to run the workflow once. The test run is independent of its normal trigger, so you can confirm each Agent node produces the expected output before it runs automatically.

   Note: The Test button appears only after you have saved the workflow changes.
3. If the workflow uses a Condition Based trigger, the Run Workflow dialog opens, since the workflow needs a specific campaign to run its test against. Select a Campaign Type and a Campaign, then click Test in the dialog to start the run.
4. The Previous run history panel opens, showing the run you just started at the top of the list.
   1. A spinner means the run is still executing.
   2. A green checkmark means the run completed successfully.
   3. A red cross means the run failed.
5. After the run completes, click View Details on that entry to review what each node produced.
6. Review the test results, and if they look correct, switch on the Publish toggle to activate the workflow. The workflow now runs automatically according to its Start trigger.

After you save the workflow, it switches from edit mode to a read-only view. In this view, the Edit and Test options appear in place of Save. You can return to edit mode by clicking Edit whenever you need to make further changes.

Build Parallel Agent Workflows

The Build a Sequential Workflow section built a sequential workflow, where each Agent node waits for the one before it to complete. This section covers the parallel pattern instead, using Duplicate and Merge nodes to run multiple Agent tasks from the same point in the workflow. For details on what these nodes do and how to configure their fields, see Duplicate Node and Merge Node in Components of a Workflow.

You can build a parallel agent workflow using the following steps.

Prerequisite:

Before branching a workflow, complete:

• Step 1: Create a Workflow
• Step 2: Define the Start Trigger
• Step 3: Add Agent Nodes

Once the workflow exists with its trigger and at least one Agent node, use the following steps to branch it into parallel paths:

Branch Node Output Using a Duplicate Node

A Duplicate node splits one node's output into two or more parallel branches, so you can run more than one Agent task from the same point in the workflow.

To add a Duplicate node wherever you want the workflow to split into multiple paths:

1. Drag a Duplicate node from the Logic section of the palette onto the canvas.
2. Connect it after the node whose output you want to branch, for example, the Start node or an Agent node.
3. In the right panel, set Number of output nodes to the number of branches you need. The default is 2.
4. Connect each of the Duplicate node's outputs to a separate node, such as an Agent node, so each branch performs its own task.

For example, to run a heatmap analysis and a session recording analysis in parallel, set Number of output nodes to 2, then connect one output to an Agent node configured for heatmap analysis and the other to an Agent node configured for session recording analysis.

Recombine Node Output Using a Merge Node

A Merge node combines the outputs of two or more parallel branches (created by a Duplicate node) back into a single output, so the workflow can continue as one path again.

To bring the parallel branches back into a single path before the workflow continues:

1. Drag a Merge node from the Logic section of the palette onto the canvas. Place it after the parallel Agent nodes you want to combine.
2. In the right panel, set Number of input nodes to match the number of branches feeding into it. The default is 2.
3. Connect the output of each parallel Agent node into one of the Merge node's inputs.
4. Connect the Merge node's single output onward to the next node, such as another Agent node or the End node.

For example, after running a heatmap analysis and a session recording analysis in parallel, set Number of input nodes to 2. Connect both Agent nodes to the Merge node, then connect its output to an End node set to Notify, so a single email summarizes both results.

Once you've connected the parallel branches to an End node, complete Step 4: Configure the Workflow's End Action and Step 5: Save, Test, and Publish the Workflow to activate the workflow.

See Manage Wingz Workflows for information on how to find a published workflow in the list, edit it, and review its run history to confirm each node completed as expected.

Troubleshooting