A Visual AI Checkpoint can be defined for each section of a Guided Journey to automatically verify that participants have reached the expected state in their VM environment.
Important Note: The improved Visual AI Checkpoint validation algorithm applies only to checkpoints whose reference screenshot was uploaded after June 24, 2026. Existing Visual AI Checkpoints that use a reference screenshot that was uploaded before this date continue to use the previous algorithm. To apply the improved algorithm to an existing checkpoint, re-upload its reference screenshot.
Note: One Visual AI Checkpoint can be created for each Guided Journey section. Creating a Checkpoint is optional, meaning that a Checkpoint can be created for some sections and not others. Visual AI Checkpoints are not available for multi-step classes.
To configure a Visual AI Checkpoint for a section in a Guided Journey:
1 Visual AI Checkpoints are integrated into a Guided Journey.
To use this feature, ensure that the Guided Journey is properly configured, as described in Guided Journeys.
2 Open the Guided Journey in the Editor.
Navigate to a section of a chapter in which you want to create a Visual AI Checkpoint. For example, as shown below:
3 Select the Checkpoint Image tab, as shown below:
4 Prepare a reference screenshot.
It is essential to provide a high-quality image for CloudShare to compare with the participant’s VM screen because the quality of this reference image directly affects the accuracy of the Visual Checkpoint AI validation mechanism.
- Recommended Method: The easiest and most recommended method for taking a screenshot is using the Take Screenshot button in the Viewer so that the reference image reflects an actual VM screen, as described in Taking a Screenshot, below.
- Best Practices: Review the Pro Tips for Accurate Matching on the right side of the screen, as shown below:
For more tips and tricks on creating Visual AI Checkpoints, see Best Practices for Visual AI Checkpoints. A link to this help center article is also provided at the bottom of the upload screen.
The following are the most important points.
- Supported Format: .png (RGBA), which provides the highest resolution for algorithm accuracy.
- Maximum File Size: 5 MB.
- Originals Only: Use full, uncropped screenshots without manual resizing.
- Match the Participant View: Make sure your screen resolution matches the participant VM environment. A Checkpoint may fail if the participant is working on a different machine, screen, or product view than the one used to create the Checkpoint.
- Stay Consistent: The more your source image reflects the live environment, the better the AI performs. Before going live, test the Checkpoint as a participant would and update the reference image or the selected area if the result is inaccurate.
5 Upload a Reference Screenshot
Upload one reference screenshot file from your computer, or drag and drop it into the upload area, shown below:
After the image is uploaded, it is displayed in the Editor while preserving the original aspect ratio. For example, as shown below:
6 Mark the Area to be Compared
Mark at least one clear, static, and unique area in the Checkpoint image that confirms the participant has reached the expected end state. For example, select a specific tab, button, field, table row, progress bar, text box, status indicator, or success message.
Only the marked areas are compared with the participant’s screen. These selected areas determine whether the Visual AI Checkpoint passes or fails.
Avoid areas that may change, such as blinking cursors or system clocks.
The comparison is not case-sensitive.
Keep Checkpoint Relevant
Each Checkpoint should directly match the objective of the section. For example, if participants are asked to configure a firewall rule, the Checkpoint should validate a visual indicator that confirms the rule was configured successfully.
For example, if the first tab must be Windows 2016 Standard, mark the first tab, as shown below:
In this case, if the first tab above reads Windows 2017 Standard, it fails.
You can edit or delete a marked area later by clicking the ellipsis menu button, as shown below:
Error messages may be displayed at the bottom of the screen:
7 Configure Failure Options.
Set how the system responds when a participant fails the Checkpoint, meaning when their screen does not show the same information as marked in the uploaded reference screenshot.
The following settings configure how participants are automatically supported when they fail a section, helping them understand their mistakes and improve their performance.
The Show Visual AI Hint and Section Hint options operate independently:
Show Visual AI Hint
Check this checkbox to specify that when a Visual AI Checkpoint fails, the incorrect areas on the participant's screen are highlighted. These highlights indicate the area of the screen that the participant must fix. These highlights are only displayed after a Checkpoint fails.
The following shows a failed Checkpoint indicating the marked area (in red) where the discrepancy was detected between the participant’s VM screen and the reference image that was uploaded when the Visual AI Checkpoint was defined.
Section Hint
Displays a custom message to participants when a Visual AI Checkpoint fails. Enter the hint text that will help participants understand what to correct and complete the section successfully. The following example shows a failed Checkpoint, which automatically displayed a hint “Try again...”.
8 Save.
Click the Save button on the bottom right of the screen to create this Visual AI Checkpoint.
Note: Only one Checkpoint can be defined per section. Checkpoints are not available for multi-step classes, and there is no limit to the number of check attempts per participant.
Click the Save button in the top right corner of the screen so that participants interact with them directly in the Viewer.
Taking a Screenshot from the Viewer
Overview
Using the Take Screenshot button in the Viewer is the recommended way to capture reference images. It captures a clean, high-quality VM screenshot at the correct resolution, without CloudShare UI elements, browser chrome, or the mouse pointer. This improves the accuracy of the Visual AI Checkpoint algorithm.
Taking a Screenshot
1 In the Instructor Console of the relevant experience, click the Instructor Environments button.
The My Environment tab shows the instructor's environment(s) and lists the VMs that it includes.
2 Click to spin up the relevant VM in that environment, as shown below:
The instructor's view is displayed, emulating the participant’s experience.
3 Start performing the steps of the Guided Journey as you would want a participant to do it.
4 After performing all the steps of a section that a participant should do, click the Take Screenshot button under the Machine section, as shown below:
5 Draw a boundary around the area to be captured in the screenshot.
The perimeter of the captured area is highlighted for a moment while the rest of the screen dims.
The VM content is captured at maximum resolution in PNG format. The browser’s Save File window is displayed. For example, as shown below:
6 Save the file to your computer.
7 Upload this screenshot file as the reference image for your Visual AI Checkpoint.
Editing a Visual AI Checkpoint While an Experience Is Live
If you save changes to a Visual AI Checkpoint while an experience is currently running, participants' content will update in real time. You will be prompted to choose how to handle existing participant progress:
- Keep sections' progress and Checkpoint results: Retains all existing participant progress and Checkpoint results. Changes apply going forward.
- Reset progress and Checkpoints for all sections: Clears all participant progress and checkpoint results for the entire Guided Journey. A fresh start from the beginning.
- Reset starting from a specific section: Resets progress and results from a chosen section onward, leaving earlier sections untouched.