Add an adaptive desktop action tool to an AI agent for web-based tasks

Configure and add a desktop action as a tool to an AI agent in AI Agent Studio so that AI agents can perform dynamic steps in the web environment.

Before you begin

Familiarize yourself with adaptive desktop actions. For more information, see Adaptive desktop actions for web-based tasks.

Role required: sn_aia.admin

About this task

Desktop actions are tools that AI agents use to interact with web applications through a browser extension. When you configure an AI agent and select desktop action as a tool, you define whether the AI agent follows a defined path (fixed steps designed in AI Desktop Actions) or an adaptive path (high-level goal described in the tool configuration). Unlike defined path desktop actions that follow preconfigured workflows, adaptive path desktop actions enable AI agents to dynamically plan and execute tasks based on high-level instructions.

An AI agent named Web Automation Agent and agentic workflow named Web Automation are provided by default when you install AI Desktop Actions. You can create a different AI agent following this procedure.

Procedure

Navigate to All > AI Agent Studio > Create and manage.
Select AI agents tab on the page.
Open the AI agent that you want to add a desktop action to.

You can select the AI agent that is provided by default (Web Automation Agent) or create an AI agent.

For creating an AI agent, see Create an AI agent.
Navigate to the Add tools and information step.
In the Add tool drop-down list, select Desktop action.

Image omitted: create-web-tool4.png
Add tool drop-down showing desktop action as an option.

On the Add a desktop action modal, select the Click here to create a desktop action link.

Image omitted: create-web-tool5.png
Add web-based desktop actions

Select the Let AI determine the steps dynamically for web-based tasks option.

Image omitted: create-web-tool6.png
Add web-based desktop actions

Select Continue
On the form, fill in the fields.

Field	Description
Name	Provide a unique intuitive name.
Tool Description	Description of the tool's purpose, functionality, inputs, and expected outputs, written in complete sentences. The AI agent uses this description to select the appropriate tool. Explain how an AI agent uses the tool and its inputs — including specific fields or data types — to carry out its role. Include exact input requirements such as format rules, character limits, and valid values. Specify what the tool returns and how the AI agent should use the output.
Navigation actions	Add a list of precise steps that the AI agent must execute on the web page or application effectively.
Time out	Enter the maximum number of minutes an AI agent should use the web page or application. Default value: 30 mins.
Execution mode	Mode of execution for your selected desktop action:- Supervised: Inputs from your live agent are required during the execution of this desktop action while the AI agent runs. - Autonomous: Doesn't require any input from your live agent during the execution of this desktop action while the AI agent runs.
Display output	Permission to display the output of the execution in the Now Assist panel or in Virtual Agent:- Yes - No If you want the AI agent to work in Off Glide architecture with AI-native experience, you must turn-on the Display output toggle. When the toggle is turned-on, you can add widgets that can be used in assistants built with AI-native experiences. The widget configuration includes: Widgets: Defines the display output to render the content in a better user experience. You can select the widget from the drop-down. Require widget transformation: An additional LLM call is required to transform the raw tool. If you choose to skip this transformation step, the tool output will be directly mapped to the widget. Yes No Display refined widget message: Refines the widget message when configured. Yes No Note: The display output as a toggle is exclusively available for the AI-native experience when the Off Glide Conversation Server plugin (com.glide.cs.offglide) is installed. If the plugin is not installed, you will continue to access the standard display output options.
Select an output transformation format	Style for the LLM to present the results as it passes information between tools and to other agents. Out transformation formats:- None - Concise - Paragraph - Summary - Custom
Write processing messages for users	Message to display to users during tool execution.- In-progress message: Write an in-progress message to be displayed to end-users while the tool is running. - Completion message: Write a completion message to be displayed to end-users once the tool finishes running.

Select Add desktop action.

The desktop action is added in the Desktop actions list on the Add tools and information page.

What to do next

For more information about executing desktop actions, see Examples of executing desktop actions using AI agents.