Business application candidate agent reference

  • Release version: Australia
  • Updated July 1, 2026
  • 3 minutes to read
    • Summarize
    Summarized using AI
    This content was generated using new OpenAI-powered functionality. Results are provided on an as is basis and are not guaranteed to be accurate or complete.

    Summary of Business application candidate agent reference

    The Business application candidate agent in ServiceNow is designed to help identify and recommend relationships between app services and business applications. It leverages AI to generate recommendations, which can be reviewed and accepted by users. This reference covers configuration properties, operational limits, underlying tables, and role requirements essential for managing and optimizing the agent's behavior.

    Show full answer Show less

    System Properties and Configuration

    All configuration properties use the prefix snsmgenai.businessappmatching. and allow you to control filtering, job scheduling, AI processing, and confidence thresholds for recommendations.

    • Scope and Filtering: Properties enable inclusion or exclusion of specific app service subclasses and allow skipping inactive app services or business applications to focus processing on relevant records.
    • Job Cadence: Controls how often synchronization and processing jobs run, with default intervals set to 7 days for data sync and hourly for processing batches.
    • Processing Tuning: Parameters like batch size (default 1000), maximum similar entries, and negative feedback entries help balance performance and AI inference quality.
    • Confidence Thresholds: Set minimum AI confidence ratings required to create recommendations and relationships, with an option to auto-accept recommendations above a certain confidence level—though manual review is the default and recommended until sufficient trust is established.
    • User Group Context: Optionally includes additional user and group context from app service records to enrich AI analysis, which is useful if varied ownership fields are populated.

    Operational Constraints

    • Maximum Supported App Services: The agent supports up to 250,000 app services. Exceeding this deactivates the synchronization job and requires contacting support for guidance.
    • Recommendation Retention: Recommendations older than two years are automatically purged to maintain system performance and storage efficiency.

    Data Tables and Role Requirements

    • The agent uses several tables for processing and storing recommendations, including snsmgenaiappserviceprocessing for job queues, snsmgenaiprocessingentry for run state data, and snsmgenaibusinessapprecommendation for recommendation records.
    • Access to these tables and agent configuration requires the itil role, which also enables users to activate/deactivate jobs, configure properties, view and manage recommendations, and create or modify business application relationships.

    Practical Use for ServiceNow Customers

    This agent empowers ServiceNow customers to automate and enhance the accuracy of linking app services to business applications using AI-driven recommendations. Proper configuration of system properties allows tailoring the agent’s behavior to fit organizational size and data characteristics. Monitoring operational limits and managing role assignments ensures stable performance and secure access. By leveraging these capabilities, customers can improve CMDB completeness, streamline impact analysis, and support better business service management decisions.

    Reference information for the Business application candidate agent, including system properties, configuration limits, table names, role requirements, and operational constraints.

    System properties

    All agent configuration properties use the prefix sn_sm_gen_ai.business_app_matching.*. You can configure the properties to control job behavior, filtering, performance, and AI confidence thresholds.

    Scope and filtering

    The following properties control which app services and business applications are included in processing.

    Table 1. Scope and filtering properties
    Property Default Description
    supported_app_service_classes (empty) Comma-separated list of app service subclasses to include. Empty (with empty unsupported list) means all classes are included.
    unsupported_app_service_classes (empty) Comma-separated list of app service subclasses to exclude.
    app_service_skip_criteria operational_status=2 Encoded query — app services that match this criterion are skipped during processing. Default skips inactive services.
    business_app_skip_criteria operational_status=2 Encoded query — business applications that match this criterion are excluded from matching. Default excludes inactive applications.
    Job cadence

    The following properties control how frequently the Data Synchronization and Processing jobs run.

    Table 2. Job cadence properties
    Property Default Description
    sync_interval_in_seconds 604800 (7 days) How often the Data Sync job re-runs after completing a full synchronization cycle.
    batch_interval_in_seconds 3600 (1 hour) How often the Processing job re-fires after completing a batch of recommendations.
    retry_interval_in_seconds 1800 (30 min) How often the Processing job polls while waiting for synchronization to finish or AI Search indexing to complete.
    Processing tuning

    The following properties tune AI inference and recommendation generation.

    Table 3. Processing tuning properties
    Property Default Description
    batch_size 1000 Number of app services processed per job execution. Hard cap: 10,000. Adjust this setting for performance on instances with very large numbers of app services.
    similar_max_entries 100 Maximum number of similar app service results returned by AI search per inference. Limits AI context and processing time.
    negative_feedback_max_entries 50 Maximum number of rejected recommendations used as negative feedback per inference. Limits the number of past rejections the AI considers when learning.
    Confidence thresholds

    The following properties control the minimum confidence levels required to generate and auto-accept recommendations.

    Table 4. Confidence threshold properties
    Property Default Description
    business_app_confidence_minimum_rating 3 (Medium) Minimum AI confidence rating (1=Very Low, 5=Very High) required to create a recommendation. Recommendations with confidence lower than this threshold are silently skipped. Increase this value to see only high-confidence recommendations; decrease to see more candidates.
    relationship_confidence_minimum_rating 3 (Medium) Minimum confidence rating required to create a CI relationship candidate when a business application match is found. Affects which related application service links are proposed.
    auto_accept_confidence_minimum_rating (empty) If set, recommendations where all confidence scores meet or exceed this value are automatically accepted without human review. Leave empty to require manual review for all recommendations.
    Tip:
    Use this setting only when you feel confident after many manual reviews.
    User group context

    The following property controls whether the AI includes additional user and group context when analyzing app services.

    Table 5. User group context property
    Property Default Description
    use_all_user_groups_info false When true, the agent gathers all user group and user values from the app service record (expanding user fields into their assignment groups) and includes them in the AI prompt. When false, only the managed_by_group value is used. Enable this property for richer context when your services have varied ownership fields populated.

    Operational constraints

    Table 6. Operational constraints
    Limit Value
    Maximum supported app services 250,000
    Recommendation retention 2 years

    If your app service count exceeds 250,000, the synchronization job deactivates itself and sets its state to unsupported. This indicates that your data volume exceeds the maximum supported scale. If this occurs, contact support for guidance on data partitioning or alternative approaches.

    Recommendations older than two years are automatically removed from the system to manage storage and improve performance.

    Table references

    The agent uses the following tables in the CMDB and internal processing system:

    Table 7. Business application candidate agent tables
    Table Name Purpose
    sn_sm_gen_ai_app_service_processing Work queue for app service processing jobs. Rows advance through states: ready → in_progress → processed.
    sn_sm_gen_ai_processing_entry Key/value run state table. Stores data_synchronization_state and per-batch processing statistics.
    sn_sm_gen_ai_business_app_recommendation Recommendation records for human review. Records are numbered with a BIZ prefix. Stores proposed business applications, confidence ratings, and related app service candidates.

    Role requirements

    Access to all Business application candidate agent tables, forms, and configuration requires the itil role. This role is required to:

    • Activate and deactivate scheduled jobs
    • Configure system properties
    • View recommendations and work queue entries
    • Accept or reject recommendations
    • Create or modify business application records and relationships