Workflow Editor
Overview
The Workflow Editor is a visual drag-and-drop designer where you build automated conversation flows step by step. Each step represents an action the system performs during a conversation — such as sending a message, asking a question, routing to a team, or invoking an AI agent. Steps are arranged sequentially, and you can add, edit, reorder, and delete them to create the exact flow your business needs.
How to Access
- Navigate to Tools > Workflows, then click on a workflow to edit it
- Required role: Admin
Interface Overview
The Workflow Editor displays:
- A visual canvas showing the workflow steps in sequential order
- A toolbox on the left with available step types organized by category
- A step editor panel on the right that appears when you click on a step
- A global settings editor accessible by clicking the workflow name or background area
- Save and Back buttons in the header
Features & Actions
Add a Step
What it does: Adds a new action to the workflow. Steps execute in the order they appear.
Steps:
- Click the + button or drag a step type from the toolbox onto the canvas
- Select the step type from the available options (see Step Types below)
- Configure the step settings in the editor panel on the right
- The step appears in the flow at the selected position
Edit a Step
What it does: Modifies the configuration of an existing step.
Steps:
- Click on the step in the visual canvas
- The editor panel opens on the right with the step's settings
- Modify the fields as needed
- Changes are applied when you save the workflow
Delete a Step
What it does: Removes a step from the workflow.
Steps:
- Click on the step in the visual canvas
- Click the delete option in the step editor or context menu
- The step is removed from the flow
Reorder Steps
What it does: Changes the order in which steps execute.
Steps:
- Drag a step up or down in the visual canvas
- Drop it in the desired position
- The workflow sequence updates accordingly
Configure Global Settings
What it does: Sets workflow-level settings that apply to the entire flow.
Steps:
- Click on the workflow background or the workflow name
- The global settings editor opens with the following fields:
- Name — The workflow name
- Inactivity Timeout — Number of days of inactivity before the session is automatically ended
- Aurora AI General Context — A text field to provide general context that all AI steps in this workflow can use
- Modify settings as needed
Save the Workflow
What it does: Saves all changes to the workflow, including steps, their configuration, and global settings.
Steps:
- Click the Save button in the header
- The workflow is saved and you receive a confirmation
Important notes:
- Always save before leaving the editor, as unsaved changes will be lost
- Saving updates the workflow definition; active sessions may continue with the previous version
Step Types
Message
What it does: Sends a text message to the customer.
Fields:
- Message — The text to send
Message with Options
What it does: Sends a message with predefined response options (buttons) for the customer to choose from. Can optionally use AI to interpret free-text responses.
Fields:
- Message — The text to display
- Options — A list of response buttons the customer can choose
- AI Intent Recognition — Toggle to enable AI-powered interpretation of free-text responses (when enabled, the AI maps the customer's message to the closest option)
Template Message
What it does: Sends a pre-configured message template (from the Templates section).
Fields:
- Template — Select from available templates
Question
What it does: Asks the customer a question and stores their response in a variable or custom property.
Fields:
- Question — The text to ask
- Variable / Custom Property — Where to store the customer's answer
Add to Team
What it does: Routes the conversation to a specific team inbox, optionally assigning a funnel stage and user.
Fields:
- Team — The target team
- Stage — Optional funnel stage to assign
- User — Optional specific user to assign
Add User
What it does: Assigns a specific user to the conversation.
Fields:
- User — The user to assign
Add Tags
What it does: Adds one or more tags to the conversation for categorization and filtering.
Fields:
- Tags — The tags to add
Go To
What it does: Jumps to a specific step in the workflow, allowing you to create loops or skip ahead.
Fields:
- Target Step — The step to jump to
AI Question
What it does: Uses AI to ask a question, interpret the response, and extract structured data.
Fields:
- Question Context — Context for the AI about what to ask
- Data Type — The type of data expected (text, number, date, etc.)
- Custom Property — Where to store the extracted value
AI Information
What it does: Uses AI to have an open-ended conversation with the customer about a specific topic, gathering information or providing assistance.
Fields:
- Initial Message — The opening message to send
- Context — Background information for the AI
- Objective — What the AI should accomplish in this step
AI Agent
What it does: Hands the conversation to a full AI agent, which takes over using all its configured instructions, knowledge base, and tools.
Fields:
- Agent — Select from available AI agents
- Include Recent History — Whether to pass the recent conversation history to the agent
- Exit on Human Input — Whether the step ends when a human agent takes over
Start Workflow
What it does: Transfers the conversation to a different workflow entirely.
Fields:
- Workflow — The target workflow to start
- Channel — The channel context for the new workflow
Fields Reference
| Field | Description | Required | Notes |
|---|---|---|---|
| Name | Workflow name | Yes | Set in global settings |
| Inactivity Timeout | Days before auto-ending idle sessions | No | Numeric, in days |
| Aurora AI General Context | Shared context for all AI steps | No | Text field |
| Step Type | The type of action to perform | Yes | Selected when adding a step |
| Message | Text content for message steps | Yes (for message steps) | Supports text formatting |
| Options | Response buttons | Yes (for message-options steps) | List of options |
| Team | Target team for routing | Yes (for add-team steps) | Select from available teams |
| Agent | AI Agent to invoke | Yes (for aiAgent steps) | Select from configured agents |
Important Notes
- Workflows execute steps sequentially from top to bottom
- Each step type has its own configuration fields
- The AI Intent Recognition feature in "Message with Options" allows customers to type free-text responses that the AI maps to the closest option
- The "Go To" step creates loops or branches in the sequential flow
- The workflow is saved as a complete definition; partial saves are not supported
- Templates, agents, teams, and custom properties must already exist before they can be selected in steps
FAQ
Q: How many steps can a workflow have? A: There is no hard limit on the number of steps. However, keep workflows focused and manageable for the best customer experience.
Q: What is the difference between AI Question, AI Information, and AI Agent steps? A: AI Question asks a specific question and extracts structured data. AI Information has an open-ended conversation about a topic with a defined objective. AI Agent hands full control to a configured AI agent with its complete set of instructions, knowledge, and tools.
Q: Can I test a workflow before deploying it? A: You can test individual AI agents in their Test Me tab. For full workflow testing, assign the workflow to a test channel and run through it as a customer would.
Q: What happens when a customer's message doesn't match any option in a Message with Options step? A: If AI Intent Recognition is enabled, the AI will try to map the response to the closest option. If disabled or if the AI cannot determine a match, the workflow may repeat the step or follow a default path.
Q: Can workflows call other workflows? A: Yes. Use the "Start Workflow" step to transfer a conversation to a different workflow.