Using Nexthink Assist

Nexthink Assist is the unified entry point for managing digital employee experience (DEX) across the Infinity platform. Embedded in search navigation, it uses a mix of the latest large language models from OpenAI and Anthropic to help IT teams instantly detect, diagnose, and resolve DEX issues, ensuring faster insights and enhanced productivity. Nexthink Assist empowers you to:

• Extract and analyze data on device performance, user sessions and applications.

• Perform contextual searches across product documentation.

• Create IT campaigns effortlessly using natural language requests.

Nexthink Assist is an artificial intelligence-based feature and, as such, may occasionally provide information that requires verification. Nexthink recommends confirming your results to ensure accuracy while using Nexthink Assist.

Getting started with Nexthink Assist

Before you begin, ensure you have the necessary permissions to view Nexthink Assist. Refer to the Permissions section for more information.

Navigate to the main menu and select Nexthink Assist.

The Nexthink Assist window appears, allowing you to type your text into the search bar. Use the following options to get the information you need:

• Search: Type text in the search bar without hitting Enter to conduct searches across various parameters. Refer to the Using Search documentation for more information.

• Assist: Type text in the search bar and hit Enter to retrieve data and answer your questions.

Type at least 3 words and press the Enter key to activate Assist. To continue using Search in standard mode, type the name of a binary, device, user or package.

Nexthink Assist can answer any questions about Nexthink, Digital Experience and your IT environment, as well as streamline campaign creation:

• Ask about Nexthink: Get quick answers about Nexthink, providing the information and context you need to get started.

• Query data: Extract and compile data, delivering instant insights on device performance, binaries, user sessions, application details and more. Nexthink Assist intelligently queries relevant sources—including Nexthink documentation, Nexthink Infinity configuration objects (such as live dashboards, applications, and library packs), and Nexthink data via NQL—providing accurate, meaningful insights tailored to your query.

• Create: Create campaigns effortlessly with natural language requests. Simply describe what you need, and let Nexthink do the rest.

Assist provides answers to your questions using Nexthink Infinity documentation, sourcing information from both in-product resources and online documentation.

To explore the data behind Nexthink Assist responses, such as NQL queries, click the embedded links or check the Sources section at the bottom for direct access to more details.

Continue the conversation by asking follow-up questions related to the current topic. To start a conversation from scratch, click Restart conversation in the top-right corner of the Nexthink Assist window.

Data visualization with Nexthink Assist

Nexthink Assist Experimental mode provides early access to beta features. The features and documentation may be subject to change before general availability.

In Experimental mode, Nexthink Assist enables you to generate charts and graphs with ease. When you request a visualization, Assist automatically runs NQL investigations, retrieves the relevant data, and delivers a chart with contextual information and embedded sources.

Accessing Experimental mode

1. Navigate to the main menu and select Nexthink Assist.

2. Type text in the search bar and hit Enter.

3. In the Nexthink Assist window, click on the action menu icon in the top-right corner.

4. Select Experimental from the drop-down menu.

By default, Assist is in Standard mode. After switching modes, click Restart conversation to ensure optimal performance.

Creating visualizations

In Experimental mode, explicitly request charts by including keywords in your text such as:

• Visualization, visualize, chart, graph, plot, or diagram when typing a request. For example: Create a bar chart to display the number of freezes by application sorted by highest to lowest.

• Press Enter to submit your request.

Example requests:

• Plot the evolution of the DEX score for the past 30 days on a line chart .

• Create a pie chart that shows the device distribution per OS platform.

• Use a bar chart to display the number of system crashes by device model. Only include the 10 worst models.

Hover over different parts of the chart to view tooltips that display data information.

Supported visualizations

Nexthink Assist supports simple chart types, such as:

• Line charts

• Bar charts

• Pie charts

• Column charts

Case study: Troubleshooting MS Teams call quality issues within the Sales department in Boston

Imagine you work for a company with a Sales department based in Boston. The team has reported issues with MS Teams call quality, and you decide to investigate the problem using Nexthink Assist. The following sections provide guidelines and examples relevant to this use case.

Ask Assist about Nexthink Infinity

To identify the cause of the MS Teams call quality issues, ask Nexthink Assist: Help me troubleshoot MS Teams call quality issues and press Enter. Nexthink Assist reviews Nexthink documentation and uses its expert knowledge to recommend troubleshooting steps to address the problem.

For more information or detailed instructions, refer to the relevant documentation pages and training materials linked by Assist at the bottom of the response.

Types of questions Assist can answer

Question examples:

• How do I configure a webhook?

• What is the difference between hard and soft navigations?

• What is the limit of campaigns I can create in Workplace Experience?

Assessing the situation with Assist

The first recommended step is to investigate the Wi-Fi signal strength.Continue the conversation with Assist in the same window by asking: Does the Boston office have a Wi-Fi strength issue? to focus on the location you suspect is experiencing poor Teams call quality. Assist conducts an assessment and confirms that the Boston office is experiencing a Wi-Fi issue.

Next, request a list of Sales department employees in Boston. You can contact these employees later via a campaign. Click the Investigations link embedded in the Assist response to access the full list. Refer to the Visual editor and NQL editor documentation for more information about investigations.

Creating campaigns with Assist

Once you have identified the problem with the Wi-Fi signal strength in the Boston office, the next step is notifying all affected employees via a Nexthink campaign.

Use Assist to create a campaign to inform employees that the Boston office has Wi-Fi issues. Assist creates a campaign—questions and answers—according to your instructions.

To improve your campaign, type additional commands into the search bar to provide further guidance.

If you are satisfied with the suggested campaign, simply ask Assist to save it as a draft in the Campaigns module, where you can then define targeted users and the preferred trigger mechanism.

Refer to the Creating Campaigns documentation for more information on how to finalize a campaign draft and distribute it among selected audiences.

Types of campaigns Assist can create

Assist can help you complete the content section for any type of Nexthink campaign. Examples:

• Create a device usage survey

• Create a campaign to inform users about ongoing Wi-Fi issues in our office

• Create an IT satisfaction survey

Refer to the Launching campaign programs documentation for guidelines on planning, gaining approval, testing and launching a campaign within your organization.

Data out of Assist scope when creating NQL queries

The Nexthink NQL data model provides an overview of the data Nexthink collects. While Assist supports most of the NQL data model, it does not currently support custom fields and the following listed fields:

Namespace	Unsupported tables
application	network_application
alert	impact
device	device/local_admins
remote_action	execution_summary
service	all
software_metering	all
connection	udp_event tcp_event
session	connect disconnect lifecycle_event lock logout unlock
web	error_summary event_summary transaction_summary page_view_summary page_view/detailed_page_load_time
workflow	execution_summary

The response times in Nexthink Assist may vary, and occasionally, some queries generated by Assist might not display correctly in Visual editor.

Frequently Asked Questions

I'm getting unexpected answers from Assist. What can I do?

It may be that the answer to your question is based on a topic Nexthink Assist does not support. Refer to the What type of Investigations can Assist create? section for more details.

Assist uses natural language to answer your questions. Sometimes, you might observe that Assist cannot respond to your question or that the response is incorrect. Here are some recommendations for how you can best phrase your questions:

• Be clear and specific: Provide clear information about the issue you are investigating. Avoid vague or ambiguous questions that may lead to inaccurate results.

• Use targeted keywords: Use relevant keywords related to the problem you are trying to solve whenever possible. This helps Assist in understanding the context and providing more accurate responses.

• Frame your questions: Structure your prompts as questions or specific commands. For example, instead of typing Find errors in Teams, try How many Teams calls had bad quality in the last 24 hours?

• Experiment and iterate: If you do not get the desired results initially, do not hesitate to refine your prompts. Assist learns and improves over time, so experimenting with different phrasing can help you achieve better outcomes.

For more information on effective prompt writing, complete the online course How to fast-track work with Nexthink Assist available on Nexthink Learn.

How can I share feedback on Assist with the Nexthink team?

Help us improve Assist by sharing your feedback using the thumbs-up and thumbs-down icons located in the lower-right corner of the latest response from Assist.

As Nexthink admin, can I pilot the usage of Assist before exposing it to a broader user group?

Yes, Nexthink admins can pilot the usage of Assist by selectively disabling and enabling the Assist permissions of certain roles.

Does Assist apply View domain constraints and permissions?

Yes, View domain applies to Assist. For example, Nexthink users querying data with Assist get the same results as running the query in NQL or the Visual editor.

Nexthink also enforces permissions before performing a task. For example, to view and use the Create campaigns feature in Nexthink Assist, ensure that your role has permission to edit campaigns. Without this permission, you cannot create campaigns and will receive the following error message: You do not have the right permissions to perform this action. Please reach out to your administrator.

Is Nexthink Assist an add-on product? Does it require an additional license?

Nexthink Assist is an Infinity feature, part of the Workplace Experience core product, and available at no additional charge.

How does Assist execute the natural language commands?

Assist executes natural language commands in three steps:

• First, when a user enters a question like "Who has been having the highest number of issues with Salesforce lightning?", the query is securely sent to OpenAI (learn more). This step helps analyze user intent and determine the most appropriate Assist tool(s)—such as an NQL query, Nexthink documentation, or campaign creation—to provide the best possible response.

• Second, one or more tools are run to gather data relevant to the user’s question. For instance, if needed, Assist leverages an internally hosted AI model to generate an NQL query, execute it within the customer's environment, and retrieve the results. This entire process takes place securely within the Nexthink infrastructure.

• Finally, an internally hosted LLM consolidates the results from the selected tool(s) into a clear and meaningful response. To ensure transparency and further exploration, links to data sources (such as NQL investigations) are included in the reply.