The Start a workflow tool allows the AI Agent to trigger a predefined workflow process from a selected application. When configured, the agent can initiate approval workflows or other automated business processes through natural language instructions.
Accessing the Tools Page
Open the target AI Agent.
Click Tools in the top navigation.
View, enable/disable, edit, or delete existing tools.
Click Add a tool to create a new tool.
Adding the “Start a workflow” Tool
After clicking Add a tool, select:
Start a workflow — Trigger a predefined workflow process
You will then configure:
Choose application – select the application that contains the workflow
Approval form / Workflow form – choose the workflow you want the agent to start
Name – the default name appears automatically but can be customized
You may click:
Add – to create the tool with default settings
Add and configure – to continue defining inputs and outputs
General Settings
As with all AI Agent tools, the following sections appear:
Detail – Name, Description, Tool type, Application, Workflow form
Credentials to use – End user credentials or Specific user
These settings behave the same across all tools.
For more details, refer to: AI agent.
Inputs – Optional fields used to prefill workflow data
Workflows may contain multiple form fields, and you can allow the AI Agent to prefill some of them when initiating the workflow.
By default, the Inputs section is empty.
To add inputs, click + Add input.
Select inputs
A popup will appear listing all available workflow form fields, including:
Input name – the field name in the workflow form
ID – internal field ID
Input type – field type (Single line, Number, Date, Option, etc.)
Check the fields you want the AI Agent to fill automatically, then click OK.
Inputs table configuration
Each selected field will appear in the table with the following columns:
Column | Description |
Input name | Workflow form field to prefill |
Input type | Field type inherited from the workflow form |
Fill using | Choose AI-generated input or Custom value |
Value | Required only when using a Custom value |
Notes:
Inputs are optional. A workflow can be started even if no fields are prefilled.
Only fields added to Inputs can be filled by the agent.
You may remove any input at any time using the delete icon.
This flexibility allows you to decide whether the workflow starts with partial data, full data, or no prefilling at all.
Note：
To learn more about the Dynamically fill with AI and Custom value options, please refer to the following article: Usage and Differences Between “Dynamically fill with AI” and “Custom value” in AI Agent Tool Inputs
Completion
Completion defines how the AI Agent behaves after triggering a workflow.
After running
You can choose one of the following options:
Waiting for response
The AI Agent waits for a response returned from the workflow.
This option is typically used when the workflow includes a Respond to AI node.
When the workflow reaches the Respond to AI node, it sends the specified data back to the AI Agent, allowing the agent to continue reasoning or respond to the user based on the workflow result.Do not response
The AI Agent does not wait for any response after starting the workflow.
The workflow continues running in the background, and the agent proceeds immediately.
This option is suitable for long-running workflows or when no immediate feedback from the workflow is required.
Relationship with Respond to AI
The Respond to AI node in a workflow determines whether and when data is returned to the AI Agent:
When Waiting for response is selected, the workflow should include a Respond to AI node to return results to the agent.
When Do not response is selected, the workflow does not need to include a Respond to AI node, and any returned data will be ignored.
For more details about the Respond to AI Action, please refer to the official documentation:Respond to AI Action(Workflow)
With this design, a workflow can function both as:
A full business process engine, and
An on-demand execution or decision service for AI Agents.
Outputs available to the agent and other tools
This section defines which outputs will be available to the AI Agent and other tools after the workflow is triggered.
For the Start a workflow tool, no outputs are added by default.
To provide information back to the agent—such as workflow instance data or execution messages—you must manually click Add output.
Managing outputs
Click Add output to select from supported output types.
Each added output appears with:
Output name
Output type
Value (Customize)
You may delete outputs that are not needed using the delete icon.
Deleted outputs can be added back at any time.
This configuration allows you full control over what information the agent receives after initiating a workflow.
Summary
The Start a workflow tool allows AI Agents to trigger approval processes or automated workflows within an application. By configuring optional inputs, you can control which workflow form fields are prefilled. Outputs provide confirmation and workflow identifiers that the agent can use in follow-up actions.
This tool is essential for workflow automation scenarios such as:
Submitting approval requests
Starting procurement or onboarding workflows
Launching automated multi-step business processes