Building Workflows

Use the AI Workflow Studio to build workflows using the building blocks, called Nodes. This section covers how to build workflows with no code or low code (using JavaScript).

Create a New Workflow

Create new workflows from the AI Workflows Studio page.

To Create a Workflow:

1. Choose AI Automation > AI Workflow Studio from the left navigation menu of the Aisera Administration application (Aisera Admin UI).

1. Click the + New Workflow button.

1. Create a Name for your workflow.

2. Select the Scope (Application/Bot or Tenant)

3. Enter a Description that describes the purpose or result of the workflow. If you're using this workflow with a v1.0 application/bot, the description is optional. If you're using this workflow with a v2.0 application/bot, this field is critical. This will be used by the LLM to determine the application/bot user's Intent. Example: Reset Windows Password.

4. Click OK.

The AI Workflow Studio canvas for you new workflow is displayed.

It includes a Success message at the top, to validate that you workflow was created successfully. Most of the Studio is the Canvas, where you can drag and drop Nodes - the components of your workflow.

After you drop a Node onto the Palette, a configuration window appears with the available options for that Node.

Building Blocks - Workflow Nodes

Nodes are the building blocks used to build the workflow and can be found on the left pane of the visual flow studio. To use any node, select the particular node, drag and drop on the editor canvas.

You can duplicate nodes by pressing ALT (or the option button on a Mac) and then selecting a node and dragging it onto the canvas.

Details of each node are available here.

Node Order

You can't add nodes to the workflow in random order and then link them together. You can only link a new node to a previous corresponding node. Connect the nodes from each category in this order:

• User Interaction

  • Message, then User Input or Input Form

• Operations

• Actions

  • Advanced (used within Action Nodes)

  • API calls can also be made from within Action Nodes.

Compatible Nodes

When you drag a node onto the Canvas, a blue line will appear if it is compatible with the previous node. When you see the blue line, drop the node onto the canvas and you node will connect to the existing workflow.

In the example above, the Message Node has been moved onto the Canvas, and will attach to the Action Node above, after it is dropped onto the canvas.

Save changes

Once your workflow logic is built, click the Save button. This saves your workflow in Draft mode.

Create Your First Workflow

This section takes you through the steps for creating a workflow that allows your application/bot users to order pizza. For simplicity, we'll use an example of a simple pizza place where customers can only order 'personal size' pizzas - one at a type. They only make 3 flavors.

Open the AI Workflow Studio by choosing Settings > AI Automation > AI Workflow Studio in the Aisera Admin UI.

This opens the 'drag-and-drop' interface, shown above, that lets you pull over sections of the workflow and configure them as you build a sequence for your application/bot to follow.

Add a Few Nodes

To begin:

1. In the upper-right section of the AI Workflow Studio window, click the + New Workflow button.

2. Enter a Name for your new workflow.

3. Choose whether you want the workflow to apply to a single application/bot or to all applications/bots created within the same tenant environment. The select Application or Tenant in the Scope pull-down menu.

   
4. Enter a Description for your workflow. Remember that the description is the purpose or result of the workflow. If you're using this workflow with a v1.0 application/bot, the description is optional. If you're using this workflow with a v2.0 application/bot, this field is critical. This will be used by the LLM to determine the application/bot user's Intent.

Message Node

1. Drag a Message Node onto the AI Workflow Studio Canvas for your new workflow.

1. This action opens the Message Node window, so you can add text. If this is the first message in your workflow, it will act as the initial message in the application/bot.

Add an introductory message or question for your application/bot user. In this example, we're going to ask the user if they want to order pizza. But what we really want to know is what type of pizza they want to order (we assume they came to our bot to order pizza).

Input Node

Since we need an answer from the user about what type of pizza they want, we will add a User Input node next.

1. Drag a User Input Node onto your workflow canvas.

2. For this use case (pizza ordering), choose Options List as the Input Type.

1. Add Value and Label options (key value pairs) at the top of the window. These will be the choices that the application/bot users can choose between.

2. Add a Message that you want to display with the choices.

3. Create an Output Variable to store the user's choice.

4. Click the OK button.

5. In order to test the workflow so far, add another Message Node at the end.

1. Now your workflow should look like this:

Test the Workflow

1. Click the Save button and then the Test button at the top of the canvas.

The Test button runs your workflow. It is a useful tool while you are developing your application/bot sequence. The first part of the workflow runs and stops to wait for the User Input. The Input Display option is set to Buttons in our example above, so the user will see the three pizza options as buttons.

After you choose an option, the workflow will proceed to the last message.

This shows you that your workflow choices are working as expected, but so far nothing is going on behind-the-scenes. In other words, nothing is happening after you choose a pizza type - no order is being placed and no pizza is actually on it's way. So now you have to program what happens after the user decides on a pizza.

Decision Node

1. Drag a Decision Node onto the canvas and put it after the User Input Node. This will automatically render as a node with your choices attached. You will also see a Default Node below the Decision Node, where you can set a default value for the pizza choice.

1. Under each leaf decision branch, drag a new user input node with a message providing the pricing for the pizza type. Each new User Input Node should look like this following.

1. Set the Input Type for each new node to Confirmation.

2. Set the output parameter value to orderDecision. When you're done with all 3 User Input Nodes, your graph will look like this:

1. In addition, gather the order decision as the user input value.

2. A decision node is required under each user input node to evaluate the value of the order decision variable. Based on this a custom message will be displayed. Drag and drop the Aisera Response nodes to each leaf branch.

Import Workflows from the Workflow Library

Browse out-of-box workflows and add them to your application. The Workflows Library can be accessed from the AI Studio Workflows page.

Export and import across applications

Import Use this option to select the workflows file to import.

Export Select this option to export the selected workflows and use it in another application or tenant.

Generate Workflows from a Knowledge Base

You can generate new workflows automatically from your Knowledge Base articles.

Refer to this document for details.

Set Message Node for Copilot Markdown Variable

You can now use a Workflow Message Node to designate that the value of a variable is in Markdown format.

To designate that the value of a variable for a Copilot application is in Markdown format:

1. Create a new application/bot or open an existing application/bot in the Aisera Admin UI.

2. Select AI Automation > AI Workflow Studio.

3. Drag a Message node onto the Workflow Studio canvas.

4. Specify that the value of the variable is in Markdown format, as shown below: