2026 May Release

Artificial IntelligencePermanent link for this heading

The following chapters describe the configuration options for AI-supported use cases.

Note:

Define AI ConfigurationsPermanent link for this heading

The AI configuration (organization > “Advanced Settings” > “AI Settings”) is needed for following use cases:

  • Provide “Ask Questions”
  • Provide “Summarize”

SettingsPermanent link for this heading

You can define following settings:

“AI Configuration” tab

  • Name
    The name of the AI configuration.
  • Endpoint
    Defines the endpoint of your Mindbreeze AI instance. The endpoint defines the index in which the data of the indexed objects are stored. If no endpoint is available for selection, please contact Fabasphere Support.
    Note: Define only one AI configuration per endpoint.
  • AI Indexing Configuration
    Defines the AI indexing configuration that determines the objects to be indexed (see chapter “Define AI Indexing Configurations”).
  • Maximum Content Size (MiB)
    Defines the maximum size of content to be included in the index.
  • Index All Objects
    Specifies whether all objects (with and without solution context) are indexed (according to the AI indexing configuration).
  • Index Objects From Following Solution Contexts
    The objects in the specified solutions are indexed (according to the AI indexing configuration).
  • Index Objects Without a Solution Context
    Defines whether objects that are not in a solution context (e.g., objects in Teamrooms) should be indexed (according to the AI indexing configuration).
  • Only Index Objects From Teamrooms in Which the AI Configuration Is Referenced
    Defines whether indexing is only performed (according to the AI indexing configuration) if the AI configuration is referenced in an app configuration, app room or Teamroom. If this option is enabled, the other indexing options are emptied.
  • Automatic Delta Indexing
    Shows whether automatic delta indexing is enabled.
  • Automatic Delta Indexing Paused
    Shows whether automatic delta indexing is paused.
  • Organization
    Shows the organization for which the configuration has been defined.

“AI Status” tab

  • AI Status
    Displays the status of Mindbreeze AI. By default, the status is updated approximately every 10 minutes.
  • AI Status Configuration
    • Status Update Interval (in Minutes)
      Defines the interval for status updates (default: 10 minutes).
    • Notified Users
      Defines the users who will be notified when the status changes from “OK” to “Not OK” or vice versa.

ActionsPermanent link for this heading

  • Enable/Disable Delta Indexing
    Defines whether changes are transferred to the index defined in the endpoint.
    Note: If delta indexing is disabled, any changes are not logged and are therefore not reflected in the index even after enabling.
  • Continue/Pause Delta Indexing
    Defines whether changes are transferred to the index defined in the endpoint.
    Note: If delta indexing is paused, any changes are logged and are thus reflected in the index after resuming.
  • Index Fully in Background
    Starts a full indexing in the background.
    Note: Run the full indexing to index existing objects. A full indexing is also useful if you have made changes to the AI configuration or AI indexing configuration.
  • Define as Default
    If no AI configuration has been explicitly defined in AI use cases, the default configuration is used. The action is only available for non-default configurations. The first AI configuration created is automatically set as the default.

Note:

  • AI configurations created in a room/app configuration context do not allow full indexing and cannot be defined as default.
  • For app configurations, app rooms, and Teamrooms, if an AI configuration is available, users with full control can use the “Tools” > “Index in Background” context menu command, where the desired AI configurations for indexing can be selected.
  • For app configurations, app rooms, and Teamrooms, users with full control can use the “Tools” > “Determine Status” context menu command. You can determine the indexed objects and the objects to be indexed.
    Note: If complex indexing settings with categories have been defined in an AI indexing configuration relevant to the current Teamroom, the number of objects to be indexed is not determined for performance reasons.

Define Insight App MappingsPermanent link for this heading

The AI configuration to be used can be specified in the properties of insight apps. If no AI configuration is specified, a mapping can be specified as defined below. If also no insight app mapping is defined, the AI configuration defined as default is used.

If necessary, define the insight app mappings in your organization under “Advanced Settings” > “AI Settings” (action “Define Insight App Mappings”):

  • Insight App
    Select the desired Insight app (e.g., “AI Chat”).
  • AI Configuration
    Defines the AI configuration to be used. The endpoint and thus the index to be used are determined from the AI configuration.

Note: The setting can also be made at room/app configuration level in the properties (“AI Settings” tab).

Evaluation logic:

  • If no entry is found at room level, an entry is determined at app configuration level (if present).
  • If no entry is found at room/app configuration level, an entry is determined at organization level.
  • If no applicable entry is found at organization level either, the AI configuration defined as default is used.

Define AI Indexing ConfigurationsPermanent link for this heading

You can use the AI indexing configuration (organization > “Advanced Settings” > “AI Settings”) to define which objects are to be indexed.

Note: The desired AI indexing configuration must be entered in the corresponding AI configuration.

  • AI Indexing Configuration
    Defines the object classes or object aspects that are to be used for indexing (e.g., “Microsoft Word Document”). The object class hierarchy is taken into account (e.g., if “Document” is selected, all content objects are indexed).
    • If a category (release version) is also specified, only objects that have both the object class/object aspect and the category (draft or release version) are indexed.
    • If AI entity definitions are also specified, these are taken into account during indexing (see chapter “Define AI Entity Definitions”).
  • Additional AI Indexing Configurations
    The indexing settings of the specified AI indexing configurations are also taken into account (including the entire configuration hierarchy).
  • Use PDF Content
    Defines whether the PDF overview should be used for indexing instead of the original content. This improves the performance of Mindbreeze AI.
  • Use AI Content
    Defines whether an additional content property should be indexed for AI use.
  • Content Property for AI Indexing
    Defines the property that should be additionally indexed for AI use.

Note:

  • All matching entries in the entire AI indexing configuration hierarchy are considered. For example, if the AI entity definitions “Ent1” and “Ent2” are defined for “Microsoft Word Document” and the AI entity definition “Ent3” is defined for “Document,” all three AI entity definitions are taken into account for Microsoft Word documents, and the AI entity definition “Ent3” is taken into account for Microsoft Excel worksheets (since it is derived from “Document”).
  • The Ready for AI Use field of objects (“General” tab) shows whether the current version of an object has already been prepared for AI use.

Define AI Entity DefinitionsPermanent link for this heading

AI entity definitions are used to provide solution-specific AI functionality. For more information, see the documentation for the respective solution.

Configuration LevelsPermanent link for this heading

AI configurations, AI indexing configurations and AI entity definitions can be defined at the following levels:

  • organization (“Advanced Settings” > “AI Settings”)
    Applies to objects that are assigned to the organization.
  • app configuration (“AI Settings”)
    Applies to app rooms and the objects they contain that are assigned to the app configuration.
  • Teamroom/app room (“Templates and Presettings” > “AI Settings”)
    Applies to objects that are assigned to the Teamroom or app room.

AI configurations are evaluated in the context in which they are stored. Make sure that the Index Objects From Following Solution Contexts and the Index Objects Without a Solution Context fields are defined corresponding to the usage if you have not enabled the Index All Objects or the Only Index Objects From Teamrooms in Which the AI Configuration Is Referenced field.

You can also store the same AI configuration in different contexts so that you do not have to define the settings multiple times.

Provide “Ask Questions”Permanent link for this heading

Mindbreeze AI's generative AI can be used to answer questions about documents, files and Teamrooms.

To use this functionality, you need:

  • An AI configuration with a defined endpoint and AI indexing configuration (see chapter “Define AI Configurations”).
  • If necessary, an Insight app mapping for the “AI Chat” app (if you do not want to use the first AI configuration stored in the relevant context or the default AI configuration).
  • The enabling of “Ask Questions” in the respective context.
    The Enable Mindbreeze AI and Display "Ask Questions" Action options can be enabled for app configurations, app rooms or Teamrooms (“AI Settings” tab). When enabling Mindbreeze AI, the default AI configuration and AI glossaries of the organization are referenced in the current context (“AI Settings” widget) and indexing is started in the background. When disabling again, the default AI configuration and glossaries are removed.
  • If necessary, a restriction to specific folders or Teamrooms (“AI Settings” tab).
    Note:
    • If a restriction is defined, only documents that are stored directly in the specified folders or Teamrooms are taken into account. Subfolders are ignored.
    • Teamrooms and folders that are not assigned to the current Teamroom can also be specified. This means that questions can be asked across Teamrooms. “Ask questions” must be enabled in all affected Teamrooms.
  • If necessary, a description of the Teamroom to provide contextual information for better answers. You can define the Description for AI using the “Define Description” action. If no Description for AI has been defined, the Description is used as a fallback. The title is always taken into account.
    Note: A maximum of 2000 characters are submitted to the AI.
  • If necessary, AI glossaries that explain relevant terms can be defined. AI glossaries of the organization have to be stored explicitly in the current context in the “AI Settings” widget. AI glossaries of app configurations are automatically considered in the assigned app rooms.
    Note: A maximum of 20000 characters are submitted to the AI.

Note: To be able to define the AI settings, you have to be authorized via the “Manage AI Settings” or “Edit Insight Apps and AI Settings” policy.

Provide “Summarize”Permanent link for this heading

Mindbreeze AI's generative AI can be used to summarize documents, files and Teamrooms.

To use this functionality, you need:

  • An AI configuration with a defined endpoint and AI indexing configuration (see chapter “Define AI Configurations”). The first AI configuration stored in the relevant context or the default AI configuration is used.
  • Perspectives, which you can define and release for usage under “Templates and Presettings” in the AI Elements area. Perspectives are questions that determine the topics for the summary. Perspectives can be grouped into perspective collections for easier selection.
    Note: Perspectives can also be defined in collections for templates and presettings for general use.
  • The enabling of “Summarize” in the respective context.
    The Enable Mindbreeze AI and Display "Summarize" Action options can be enabled for app configurations, app rooms or Teamrooms (“AI Settings” tab). When enabling Mindbreeze AI, the default AI configuration and AI glossaries of the organization are referenced in the current context (“AI Settings” widget) and indexing is started in the background. When disabling again, the default AI configuration and glossaries are removed.
  • If necessary, own AI prompts for the final summary (“AI Settings” tab). If no AI prompts are defined, the provided AI prompts “Default Prompt for Summary (Introduction)” and “Default Prompt for Summary (Instructions)” will be used. Open the properties of the default AI prompts to see an example of how to define your own AI prompts. The AI Prompt Instructions for No Answer field is not taken into account for summaries.
    Note: AI prompts can be defined in the organization or in app configurations, app rooms, and Teamrooms under “AI Settings.”
  • If necessary, a restriction to specific folders or Teamrooms (“AI Settings” tab).
    Note:
    • If a restriction is defined, only documents that are stored directly in the specified folders or Teamrooms are taken into account. Subfolders are ignored.
    • Teamrooms and folders that are not assigned to the current Teamroom can also be specified. This means that questions can be asked across Teamrooms. “Summarize” must be enabled in all affected Teamrooms.
  • If necessary, a description of the Teamroom to provide contextual information for better answers. You can define the Description for AI using the “Define Description” action. If no Description for AI has been defined, the Description is used as a fallback. The title is always taken into account.
    Note: A maximum of 2000 characters are submitted to the AI.
  • If necessary, AI glossaries that explain relevant terms can be defined. AI glossaries of the organization have to be stored explicitly in the current context in the “AI Settings” widget. AI glossaries of app configurations are automatically considered in the assigned app rooms.
    Note: A maximum of 20000 characters are submitted to the AI.

Note: To be able to define the AI settings, you have to be authorized via the “Manage AI Settings” or “Edit Insight Apps and AI Settings” policy.

Provide “Metadata Extraction”Permanent link for this heading

With the help of Mindbreeze AI, metadata can be extracted from the documents being registered and imported directly during the registration process.

To use this functionality, you need:

  • An AI configuration with a defined endpoint and AI indexing configuration (see chapter “Define AI Configurations”). The first AI configuration stored in the relevant context or the default AI configuration is used.
  • Metadata extractions and extraction definitions, which you can define and release for usage under “Templates and Presettings” in the AI Elements area.
    Note: Metadata extractions and extraction definitions can also be defined in collections for templates and presettings for general use.
  • Incoming categories intended for metadata extraction. Metadata extraction is performed only if a corresponding incoming category is assigned to the object to be registered. By default, the incoming categories “Incoming Category for Generic Registration” and “Incoming Category for Generic Registration With Form” are available with metadata extraction enabled. Solutions may provide additional incoming categories.

Note:

  • To be able to define the AI settings, you have to be authorized via the “Manage AI Settings” or “Edit Insight Apps and AI Settings” policy.
  • To be able to define the AI settings, you have to be authorized via the “Edit Forms and Categories” policy.

Metadata Extraction

The metadata extraction offers the following settings:

  • Name
    The name of the metadata extraction.
  • Applicable to
    The metadata extraction is only offered during registration if all of the following conditions are met (AND operator):
    • The object class of the object to be registered must correspond to one of the defined object classes or be derived from one of the defined object classes. If no object class is defined, all object classes are permitted.
      Example: If the metadata extraction defines the object class “Document”, it applies to Microsoft Word documents, text documents, etc.
    • The object aspect of the object to be registered must correspond to one of the defined object aspects or be derived from one of the defined object aspects. If no object aspect is defined, all object aspects are permitted.
    • The category of the object to be registered must correspond to one of the defined categories or be derived from one of the defined categories. If no category is defined, all categories are permitted.
  • Extraction Definitions
    To extract the desired values from the document to be registered, a separate extraction definition is required for each value.
    • Extraction Definition
      Defines the extraction definition (see the next section, “Extraction Definition”).
    • Target Property
      Defines the property in which the extracted value is to be stored. Do not specify a property if the value is to be extracted but not stored.
      Note: The selectable properties are determined based on the defined usability. To do this, the intersection of all properties of the defined object classes, the intersection of all properties of the defined object aspects, and the intersection of all properties of the defined categories are formed. The properties of these three intersections together form the selectable properties. If no usability is defined, the properties of the abstract base object class “Object” are available (e.g., Subject).
    • Expression for Calculating the Target Object
      If multiple objects are created during registration (e.g., a file and a content object), an app.ducx expression can be used to specify the target object for the target property. By default, the object displayed during registration is used as the target object.
      Example: ::targetobject = cooobj.FSCFUNDINGROOM@1.1001:ffmaincontent;
  • Expression for Finalizing Metadata Extraction
    Defines an app.ducx expression that is executed after metadata extraction (e.g., for validations or post-processing steps).

Extraction Definition

The extraction definition offers the following settings:

  • Name
    The name of the extraction definition.
  • Applicable to
    The extraction definition can only be used in a metadata extraction if all of the following conditions are met (AND operator):
    • For each object class listed in the metadata extraction, the corresponding object class or a base object class of that object class must be defined. Additional object classes may also be defined.
    • For each object aspect listed in the metadata extraction, the corresponding object aspect or a base object aspect of the object aspect must be defined. Additional object aspects may also be defined.
    • For each category listed in the metadata extraction, the corresponding category or a base category of the category must be defined. Additional categories may also be defined.
  • Prompt for Extracting the Value
    Defines a prompt that tells the AI how to extract the value from the document.
  • Expression for Calculating the Prompt for Extracting the Value
    Defines an app.ducx expression for dynamically determining the prompt.
    Example:
    if (::targetobject.objname contains "internal") {
      "    Extract the internal bill number";
    }
    else {
      "    Extract the bill number";
    }
    Note: Is only evaluated if no prompt for extracting the value is defined.
  • Property for Extracting the Value
    Defines a property that sets the desired prompt in the AI Description field. If no AI description is defined, the property’s context-sensitive help is used. If an AI description is available, it is displayed read-only in the AI Description From Property for Extracting the Value field.
    Note: This field is only evaluated if no prompt for value extraction and no expression for calculating the prompt for value extraction is defined.
  • Expression for Finalizing the Extraction of the Value
    Defines an app.ducx expression that is executed after the value is extracted. This allows, for example, the extracted value to be modified subsequently.
    Example: ::metadata.GetEntry("objsubject") + " (Finalization)";

AI Consumption InformationPermanent link for this heading

In your organization, under “Usage” > “AI Consumption Information,” you can find the volume of indexed documents and API calls used per endpoint.

Download PDF

Download PDF