Kodexa's document labeling system enables you to efficiently highlight, categorize, and extract information from documents. When you select text in a document, a popup appears with labeling options, chat templates, and tools to help you work with the content.

To begin labeling:

Click the Properties button to open the properties panel for the selected content. This shows detailed information about the selection including node structure and features (available to users with developer tools enabled).

Click the Copy button to copy the selected text to your clipboard for use elsewhere.

Click New Data Element to add the selected text as a new field in your Data Definition (available to users with developer tools enabled). This creates a new label that can be reused throughout your project.

Type in the search box to find labels by name. The system shows matching labels from both your Data Definition and AI models.

Use special characters to filter label results:

Example: Type `invoice to see only model labels containing "invoice"

When no search query is entered, the popup shows available message templates with an orange chat icon. These templates allow you to:

Click a message template to send it to the chat. If the template requires additional information, you'll be prompted to fill in options before sending.

The popup shows all labels currently applied to your selection, organized by:

Labels are organized into groups:

Click any label with a plus icon to apply it to your selection. The label is immediately saved and the popup closes.

To remove a label:

If you have more than 3 labels selected, a Clear Selected Labels option appears to remove all labels at once.

For users with developer tools enabled, click "View Feature Metadata" under any label to see detailed technical information about the label including:

Click the book icon next to a label to open its definition in the Data Definition panel. This shows the label's configuration, properties, and how it's used across your project.

Labels marked with a blue "Model" badge were suggested or created by AI assistants. These labels come from trained models rather than manual labeling.

Some labels show their parent label name beneath them, indicating they belong to a hierarchical structure. For example, "Invoice Number" might be under the parent "Invoice Header".

When AI models detect labels with low confidence that match your selected text, they appear in a special section. These are suggestions that:

Click the replace icon on a no confidence tag to apply it to your selection.

If a document is locked, the label popup shows only the selected text and a copy button. You cannot add or remove labels from locked documents.

When labeling in a test result (execution result), a red warning appears indicating that labels cannot be saved. This prevents accidentally labeling test data that will be discarded.

The Data Definition panel on the left side of the workspace shows the complete structure of labels available for your project. You can manage label properties, hierarchies, and configurations from this panel.