Now Assist AI agents reference

  • リリースバージョン: Australia
  • 更新日 2026年03月12日
  • 所要時間：21分

    • Find more information about user roles, tables, and the different properties that are installed in Now Assist AI agents.

    Now Assist AI agents roles

    The following roles are installed with Now Assist AI agents with a compatible Now Assist application.

    表 : 1. Roles needed to configure or monitor AI agents
    Role Description
    AI Agent Admin [sn_aia.admin] Administrator of the application. A user with the sn_aia_admin role can create, read, update, and delete records.
    AI Agent Viewer [sn_aia.viewer] Read-only access to the application. A user with the sn_aia_viewer role has read and report access on all tables.
    agent_role_config_admin With this role, user can access and modify Agent role configurations with AI Agent Admin [sn_aia_admin] being the parent role.
    agent_role_config_viewer Can view the Agent role configurations with AI Agent Viewer [sn_aia_viewer] being the parent role.

    Now Assist AI agents system properties

    The following are system properties that define default values and behavior.

    表 : 2. System properties to use for configuring AI agents
    Property Description
    sn_aia.agent_llm_provider Defines the (large language model) LLM service provider for AI agents.

    Default value: azure_openai.
    sn_aia.agent_tool_supported_data_types Defines a comma-separated list of supported data types for tools that are used by agents for IntegrationHub spoke. Each value corresponds to the name field of records in the Field classes table [sys_glide_object].
    sn_aia.analytics_dashboard_sysid Provides the sys_id for the AI Agents Analytics dashboard.

    read_roles: sn_aia.admin and sn_aia.viewer
    sn_aia.continuous_communicator_output_limit Defines the maximum number of continuous outputs that the AI Agent Orchestrator can trigger to show to users.

    Default value: 3
    sn_aia.continuous_tool_execution_limit Defines the maximum consecutive number of uses for the same tool.

    Default value: 7
    sn_aia.enable_usecase_tool_execution_mode_override Enables running agentic workflows fully autonomously, overriding any non-automated tools in the agentic workflow.

    Default value: false
    sn_aia.ltm.category.auto_create Enables AI-generated categories if no matching categories exist.

    Default value: false
    sn_aia.ltm.enable_long_term_memory Enables long-term memory for AI agents. All previous user interactions are used as context for the LLM.

    Default value: false
    sn_aia.maximum_agent_tools Defines the maximum number of tools that can be assigned to an AI agent.

    Default value: 20
    sn_aia.max_scheduled_trigger_query Defines how many records are processed when a scheduled trigger is detected.

    Default value: 10

    Write_role: admin
    sn_aia.mid_skill_switch_enabled Enables mid-skill switching.

    Default value: false
    sn_aia.react_failure_retry_max_limit Defines the maximum number of retries in case of an execution failure.

    Default value: 3
    sn_nowassist_va.router_redirect_va_agentic Determines AI agent discovery in Virtual Agent. If set to NEVER, Virtual Agent continues Q&A without any agentic AI.

    Default: ROUTER_DECISION
    com.glide.cs.dynamic.capability.timeout Defines the Timeout for AIA Proficiency Descriptor.

    Default value: 180.

    Write_role: admin.
    sn_aia.enable_follow_up Enables users to continue the conversation with follow-ups after the agentic workflow execution is complete.

    Default value: true.
    sn_aia.follow_up_message Defines a follow-up message sent after execution is completed.

    Default Value: How else can I help you?
    sn_aia.allow_context_sharing Enables the sharing of short-term memory, allowing context to persist across execution within the same conversation.

    Default Value: true
    sn_aia.agent_strategy_choice_enabled Enables to show the LLM reasoning strategy in the agent setup screen.

    Default Value: false
    sn_aia.context_sharing_strategy This property defines the strategy to use for storing short-term memory for an execution.

    Default Value: summarise
    sn_aia.enable_agent_tool_input_value_overrides Enables you to override the agent tool input value.

    Default Value: true.
    sn_aia.follow_up_qna_failure_limit Defines the limit to exit execution if the number of consecutive questions and answers aren’t available in the follow-up.

    Default value: 1.
    sn_aia.ltm.use_memory_for_ai_agent Enables long-term memory for AI agent interactions. When enabled, stored user memories are utilized in AI agent interactions.

    Default value: true.
    sn_aia.quick_mode_failure_retry_max_limit Defines maximum limit for retries in case of a failure in Quick Mode execution.

    Default value: 3.
    sn_aia.user_context_data Defines a comma-separated list of user context data to be used with AI Agents.

    The list is used to pick the data available from knowledge graph API: getUserContext.

    List of available user information:
    • profile
    • manager
    • reportees
    • assets
    The user information can also be customized by overriding the method getUserContext via UserContextUtil.

    If customized, the property must define the comma separated list of keys generated by the customized getUserContext method.

    Default value: profile.
    sn_aia.external_agents.enabled

    Set to true to enable external agent features
    sn_aia.external_agents.parallel_conversations.enabled

    Enables or disables multiple simultaneous conversations per user
    sn_mcp_client.cursor.max_iterations
    Specifies the maximum number of cursor-based pagination iterations to perform when fetching MCP tools. Helps prevent excessive looping or API calls during data retrieval.
    注:
    Set this value to 0 to disable pagination iteration limits.
    mcp_guardian_check Enables guardian check for MCP Client when the value is set to true.

    The default value is false.

    注:
    To enable guardian check for MCP Client, ensure that you enable Now Assist guardian on AI Agent Studio > Settings page.
    com.glide.agentic_processes_view.enabled Enables the in-product experience for agentic workflows in the AI Workflows panel. Ensure that this is set to true if you plan on using UI actions to run agentic workflows.

    Properties on the Agent Properties table

    The following properties are on the Agent Properties [sn_aia_property] table. These properties affect different AI agent behaviors.

    表 : 3. General properties on the Agent Properties table

    Name

    Description

    Default value
    alert.assist_spike_hours_to_check Time interval, in hours, between running the scheduled job that checks for spikes in assist usage 3
    alert.assist_spike_usage_percentage_threshold Percentage increase required from previous job results to trigger the spike in assist usage notification 0.5 (50%)
    alert.assist_spike_usage_threshold Minimum number of assists to trigger the spike in assist usage notification 5000
    alert.consecutive_error_count Consecutive number of latency errors to trigger the latency notification 3
    alert.llm_latency_threshold Time taken, in seconds, for the LLM to respond before sending a latency error 10
    alert.tool_latency_threshold Time taken, in seconds, for a tool to finish executing before sending a latency error 300
    enable_agent_tool_input_value_overrides Determines whether tool input values can be overriden by the AI agent or other tools for use by another tool false
    execution_task.latency_thresholds Determines thresholds on LLM and tool execution times and response tokens/character counts for tool calling 
    {
  "llm": {
    "timeBoundaries": [
      5000,
      10000
    ],
    "tokenLimit": 500
  },
  "tool": {
    "timeBoundaries": [
      200000,
      300000
    ],
    "outputCharLimit": 10000
  }
}
    follow_up_behaviour Actions to take by the agentic workflow after it has finished executing. Each record applies to only one agentic workflow. no_followup_close_conversation
    mcp_guardian_check Determines whether Now Assist Guardian runs on MCP tool executions false
    show_citations Determines whether agentic AI-generated responses in Now Assist panel or Now Assist in Virtual Agent add citations for their output false

    The following properties are used to detect and avoid infinite recursion or execution loops. If you reach the max executions that match the query within the time window, any new executions matching the query will abort. You can adjust these values if you want to lower the threshold for detecting recursion.

    表 : 4. Recursive check properties on the Agent Properties table
    recursive_check.create_max_executions Maximum number of matching executions creating records 50
    recursive_check.create_time_window Time window, in minutes, for checking for matching executions creating records 15
    recursive_check.query_for_create_record Query for the fields of execution plans to create records to check if executions are repeating objective={objective}^agent={agent}^ORusecase={usecase}
    recursive_check.query_for_update_record Query for the fields of execution plans to update records to check if executions are repeating related_task_record={related_task_record}^objective={objective}^agent={agent}^ORusecase={usecase}
    recursive_check.update_max_executions Maximum number of matching executions updating a record 5
    recursive_check.update_time_window

    Time window, in minutes, for checking for matching execution updates

    		 15

    Now Assist AI agents tables installed

    The following tables are installed so Now Assist AI agents works as expected:

    表 : 5. Tables used for configuring AI agents
    Table Description
    Agentic workflows [sn_aia_usecase] List of configured agentic workflows.
    AI Agents [sn_aia_agent] List of configured AI agents.
    Tools [sn_aia_tool] List of tools used by an AI agent.
    Strategies [sn_aia_strategy] List of strategies used by an AI agent.
    Teams [sn_aia_team] Team that is a group of listed agents.
    Team members [sn_aia_team_member] List of teams mapped to an agent.
    Agent Tools [sn_aia_agent_tool_m2m] List of tools mapped to an AI agent.
    AIA Trigger Configurations [sn_aia_trigger_configuration] List of triggers created for an agentic workflow.
    Execution Tasks [sn_aia_execution_task] List of tasks by execution plan ID.
    Messages [/sn_aia_message] List of messages recorded in AI agent conversations to and from the human users.
    Tools Executions [sn_aia_tools_execution] List of tools executed by the plan ID.
    注:
    The records in the Tools Executions table expire and become unavailable after a period of 13 months.
    Execution Plans [sn_aia_execution_plan] List of plan executions by conversation ID.
    Agent Tools [sn_aia_agent_tool_m2m] List of tools and maximum automatic executions.
    AI Agent configs [sn_aia_agent_config] List of active AI agents configured for the proficiency that they’ll be used in.
    Agent properties [sn_aia_property] Various properties that affect AI agent behavior.
    Gen AI Metadata M2M [sn_aia_gen_ai_m2m] List of Gen AI metadata and the maximum automatic executions.Maintains the mapping between sn_aia_execution_task and Gen AI log metadata.

    If two large language model (LLM) calls are made to the sn_aia_execution_task, then the sn_aia_gen_ai_m2m table has two records.
    Report metrics [sn_aia_report_metric] List of the report metrics.
    Agent Access Role Configurations [sys_agent_access_role_configuration] List of agent access roles. You can also create new agent access roles from this table.
    Sys_generative_ai_configuration
    表 : 6. Tables used for Group Action FrameworkSee Group Action Framework for more information.
    Table Description
    GAF record group [sn_gaf_record_group] Stores the output of the grouping skill. Each record represents a cluster of related records. You can open each record group record to discover which records were included within the cluster.
    GAF record group detail [sn_gaf_record_group_detail] Contains the individual records that belong to each group GAF record group. You can also find if a record is marked to act as a representative of the cluster on these records.
    GAF action strategy result [sn_gaf_action_strategy_result] Holds the results of the Action Strategy skill, which selects representative records from each group for downstream processing.
    GAF action mapper results [sn_gaf_action_mapper_result] Stores the output of the Mapper skill, which maps new records to existing clusters.
    GAF action reducer results [sn_gaf_action_reducer_result] Stores the result of the Action Reducer skill. The results include insights for entire clusters. For example, how to resolve incidents similar to those gathered in a cluster.