Business application candidate agent reference

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

Related tasks

• Activate the Business application candidate agent
• Review Business application candidate agent recommendations