Configuring Agentic Text Monitoring for Home Office

Configuring Agentic Text Monitoring for Home Office for

To configure this agent:

  1. Ensure Configuring Vault AI in CRM is complete.
  2. Navigate to Admin > Users & Groups > Permission Sets.
  3. Select the appropriate permission set.
  4. Select the Agents tab.
  5. Select the Execute checkbox for the Text Monitoring for Home Office Agent (text_monitoring_for_home_office_agent__v).
  6. Select the Objects tab.
  7. Grant the following permissions:

    Object

    Object Permission

    Object Types

    Fields

    Field Permission

    agent_rule__v

    CRE

    text_monitoring_agent_rule__v

    • acceptable_example__v
    • applicable_countries__v
    • applicable_fields__v
    • applicable_user_types__v
    • display_order__v
    • instruction__v
    • name__v
    • unacceptable_example__v

    Edit

    monitored_text__v

    CRE

    • agent_monitored_text__v
    • agent_flagged_phrase__v
    • issues_found__v
    • reviewer_determination__v
    • reviewer_notes__v

    Edit

    • agent_flagged_phrase__v
    • agent_reasoning__v
    • agent_rule__v
    • agent_rule_instruction__v
    • application_profile__v
    • assignee__v
    • category__v
    • complete_agent_request__v
    • complete_agent_response__v
    • country__v
    • created_by__v
    • created_date__v
    • detail_group__v
    • displayed_to_end_user__v
    • external_id__v
    • flagged_phrase__v
    • global_id__sys
    • id
    • is_parent__v
    • feedback__v
    • job_id__v
    • modified_by__v
    • modified_date__v
    • legacy_crm_id__v
    • lifecycle__v
    • stage__sys
    • state__v
    • state_stage_id__sys
    • link__sys
    • vault_id_link__v
    • monitored_date__v
    • monitored_field__v
    • monitored_object__v
    • name__v
    • monitored_text_status__v
    • object_type__v
    • monitoring_rule_name__v
    • ownerid__v
    • parent_monitored_text__v
    • product__v
    • security_profile__v
    • status__v
    • text__v
    • text_created_by__v
    • text_last_modified_by__v
    • text_last_modified_date__v
    • text_owner__v
    • user_profile__v
    • user_type__v
    • vault_id__v
    		Read
  8. Grant Edit access to the flagged_text__v field on the appropriate objects (enables flagging monitored text for records on the object):
    • Account Plan (account_plan__v)
    • Action Item (action_item__v)
    • Call Report (call2__v)
    • Call Discussion (call2_discussion__v)
    • Call Follow-Up Task (call_followup_task__v)
    • Call Objective (call_objective__v)
    • Commercial Insights (commercial_insights__v)
    • Expense (expense__v)
    • Goal (goal__v)
    • Medical Interaction (medx__v)
    • MedX Discussion (medx_discussion__v)
    • Medical Discussion (medical_discussion__v)
    • Medical Insights (medical_insight__v)
    • SWOT(swot__v)
  9. Navigate to Admin > Configuration > Objects > ai_settings__v.
  10. Select the Layouts tab.
  11. Place the home_office_text_monitoring__v field on the appropriate layouts.

Creating Agent Rule Layouts

To customize the information displayed to compliance admins, create layouts for agent rules:

  1. Navigate to Admin > Configuration > Objects > agent_rule__v.
  2. Select the Layouts tab.
  3. Create a layout for the text_monitoring_agent_rule__v object type.
  4. Place the appropriate fields on the layout, in the desired order:
  • acceptable_example__v
  • applicable_countries__v
  • applicable_fields__v
  • applicable_user_types__v
  • display_order__v
  • instruction__v
  • name__v
  • unacceptable_example__v

To configure this agent:

  1. Ensure Configuring Vault AI in CRM is complete.
  2. Navigate to Admin > Users & Groups > Permission Sets.
  3. Select the appropriate permission set.
  4. Select the Agents tab.
  5. Select the Execute checkbox for the Text Monitoring for Home Office Agent (text_monitoring_for_home_office_agent__v).
  6. Select the Objects tab.
  7. Grant the following permissions:

    Object

    Object Permission

    Object Types

    Fields

    Field Permission

    monitored_text__v

    CRE

    • agent_monitored_text__v
    • agent_flagged_phrase__v
    • issues_found__v
    • reviewer_determination__v
    • reviewer_notes__v

    Edit

    • agent_flagged_phrase__v
    • agent_reasoning__v
    • agent_rule__v
    • agent_rule_instruction__v
    • application_profile__v
    • assignee__v
    • category__v
    • complete_agent_request__v
    • complete_agent_response__v
    • country__v
    • created_by__v
    • created_date__v
    • detail_group__v
    • displayed_to_end_user__v
    • external_id__v
    • flagged_phrase__v
    • global_id__sys
    • id
    • is_parent__v
    • feedback__v
    • job_id__v
    • modified_by__v
    • modified_date__v
    • legacy_crm_id__v
    • lifecycle__v
    • stage__sys
    • state__v
    • state_stage_id__sys
    • link__sys
    • vault_id_link__v
    • monitored_date__v
    • monitored_field__v
    • monitored_object__v
    • name__v
    • monitored_text_status__v
    • object_type__v
    • monitoring_rule_name__v
    • ownerid__v
    • parent_monitored_text__v
    • product__v
    • security_profile__v
    • status__v
    • text__v
    • text_created_by__v
    • text_last_modified_by__v
    • text_last_modified_date__v
    • text_owner__v
    • user_profile__v
    • user_type__v
    • vault_id__v
    		Read
  8. Grant Edit access to the flagged_text__v field on the appropriate objects (enables flagging monitored text for records on the object):
  9. Account Plan (account_plan__v)
  10. Action Item (action_item__v)
  11. Call Report (call2__v)
  12. Call Discussion (call2_discussion__v)
  13. Call Follow-Up Task (call_followup_task__v)
  14. Call Objective (call_objective__v)
  15. Commercial Insights (commercial_insights__v)
  16. Expense (expense__v)
  17. Goal (goal__v)
  18. Medical Interaction (medx_v)
  19. MedX Discussion (medx_discussion__v)
  20. Medical Discussion (medical_discussion__v)
  21. Medical Insights (medical_insight__v)
  22. SWOT(swot__v)

Creating Monitored Text Layouts

To customize the information displayed to compliance monitors, create layouts for Agent Monitored Text and Agent Flagged Phrase object types.

Agent Monitored Text

To create this layout:

  1. Navigate to Admin > Configuration > Objects > monitored_text__v.
  2. Select the Layouts tab.
  3. Create a new layout for the agent_monitored_text__v object type.

  4. Create the following Detail Form sections:

    • Review Assignment (two-column layout)
    • Monitored Record Details (two-column layout)
    • Agent Request and Response (two-column layout)
    • Workflow Timeline, if not already on the layout (one-column layout)
  5. Create a Related Object section for Agent Flagged Phrase (s).

  6. Include the following columns in the Agent Flagged Phrase (s) section:

    • Monitored Text Name (name__v)
    • Agent Rule (agent_rule__v)
    • Agent Reasoning (agent_reasoning__v)
    • Text (text__v)
    • Reviewer Determination (reviewer_determination__v)

  7. Place the following fields in the Detail Form sections:

    • Review Assignment:
    • ID (id)
    • Created By (created_by__v)
    • Created Date (created_date__v)
    • Last Modified By (modified_by__v)
    • Last Modified Date (modified_date__v)
    • Assignee (assignee__v)
    • Issues Found (issues_found__v)
    • Status (status__v)
    • Lifecycle (lifecycle__v)
    • Lifecycle Stage (stage__sys)
    • Lifecycle State (state__v)
    • Owner (ownerid__v)
    • Monitored Record Details:
    • Link to Flagged Record (vault_id_link__v)
    • Monitored Date (monitored_date__v)
    • Monitored Object (monitored_object__v)
    • Monitored Field (monitored_field__v)
    • Text Owner (text_owner__v)
    • Text Last Modified Date (text_last_modified_date__v)
    • User Country (country__v)
    • User Profile (user_profile__v)
    • User Type (user_type__v)
    • Agent Request and Response:
    • Complete Agent Request (complete_agent_request__v)
    • Complete Agent Response (complete_agent_response__v)

    Fields do not need to be placed in the Workflow Timeline section.

  8. Select Save.

Agent Flagged Phrase

To create this layout:

  1. Navigate to Admin > Configuration > Objects > monitored_text__v.
  2. Select the Layouts tab.
  3. Create a new layout for the agent_flagged_phrase__v object type.

  4. Create the following Detail Form sections:

    • Flagged Phrase (two-column layout)
    • Agent Rule (two-column layout)
    • Review (two-column layout)
    • Workflow Timeline, if not already on the layout (one-column layout)

  5. Place the following fields in the Detail Form sections:

    • Flagged Phrase:
    • Parent Monitored Text (parent_monitored_text__v)
    • Text (text__v)
    • Flagged Phrase (flagged_phrase__v)
    • Text Owner (text_owner__v)
    • Text Created By (text_created_by__v)
    • Text Last Modified By (text_last_modified_by__v)
    • Text Last Modified Date (text_last_modified_date__v)
    • Agent Rule:
    • Agent Flagged Phrase (agent_flagged_phrase__v)
    • Agent Rule (agent_rule__v)
    • Agent Rule Instruction (agent_rule_instruction__v)
    • Agent Reasoning (agent_reasoning__v)
    • Review:
    • Reviewer Determination (reviewer_determination__v)
    • Reviewer Notes (reviewer_notes__v)

Fields do not need to be placed in the Workflow Timeline section.

Creating an Agent Monitored Text Tab

To provide a dedicated tab for compliance monitors to access Agent Monitored Text records:

  1. Navigate to Admin > Configuration > Tabs.
  2. Select Create.

  3. Populate the following fields:

    • Tab Type - Object
    • Object - monitored_text__v
    • Object Type - agent_monitored_text__v
  4. Select Save.
  5. Navigate to Admin > Users and Groups > Permission Sets.
  6. Select the appropriate permission set.
  7. Select the Tabs tab.
  8. Grant View permission to the newly created tab.

Creating a Monitored Text Workflow

Admins can create custom lifecycles and workflows on the Monitored Text object to reflect their organization’s compliance monitoring processes. These lifecycles and workflows vary depending on the organization. For more information, see Lifecycles & Workflows on the Vault platform help site.

When building a custom lifecycle and workflow on the Monitored Text object, ensure it works for all functionality relying on this object: Agentic Text Monitoring, Agentic Text Monitoring for Home Office, and Approved Notes, if enabled in the Vault.

To configure this agent:

  1. Ensure Configuring Vault AI in CRM is complete.
  2. Navigate to Admin > Users & Groups > Permission Sets.
  3. Select the appropriate permission set.
  4. Select the Agents tab.
  5. Select the Execute checkbox for the Text Monitoring for Home Office Agent (text_monitoring_for_home_office_agent__v).
  6. Navigate to Business Admin > Objects > AI Settings.
  7. Select the home_office_text_monitoring__v AI setting to On Submit, to enable the feature for the applicable users. These are the users whose submitted records will be monitored by the agent.
  8. Activate the Text Monitoring for Home Office Agent:
    • Navigate to Admin > Configuration > Vault AI Setup > Agents
    • Select the Text Monitoring for Home Office Agent (text_monitoring_for_home_office_agent__v)
    • Select Edit
    • Select Active in the Status picklist
    • Select Save
    • Expand the Actions section
    • Verify the appropriate agent actions are Active for the agent:
    • Home Office Text Monitoring - Apply Agent Rules for Home Office Text Monitoring

Creating Text Monitoring for Home Office Agent Rules as

Agent Rules determine how records are reviewed for home office monitoring. Veeva provides a set of standard rules covering common pharmaceutical compliance topics, allowing an organization to launch a text monitoring program without authoring Agent Rules from scratch. These rules are delivered inactive by default. Admins can activate a standard rule as-is, or copy a standard rule and adjust editable fields to match the organization's specific requirements.

Alternatively, admins can create custom rules. When using custom rules, admins are responsible for testing and updating rules.

At least one active rule is required to use Text Monitoring for Home Office.

Standard Rules

Standard rules and their associated Agent Rule (agent_rule__v) records are available by default. Customers can activate or inactivate the Veeva-provided Agent Rules, but they cannot delete them.

The only fields editable on Agent Rule records for standard rules are the following:

  • Applicable Fields - Field and object combinations to be monitored, in a semicolon-delimited list using dot notation. For example, to monitor the Call Text field, enter the text: call2__v.call_notes__v. Text fields on the following objects are supported:
    • Account Plan (account_plan__v)
    • Action Item (action_item__v)
    • Call Report (call2__v)
    • Call Discussion (call2_discussion__v)
    • Call Follow-Up Task (call_followup_task__v)
    • Call Objective (call_objective__v)
    • Commercial Insights (commercial_insights__v)
    • Expense (expense__v)
    • Goal (goal__v)
    • Medical Discussion (medical_discussion__v)
    • Medical Insights (medical_insight__v)
    • SWOT (swot__v)

    Only fields with data type Text, Rich Text, or Long Text are supported for monitoring.

  • Applicable Countries - Countries where the rule applies. This is matched to the User Country. If left blank, the rule applies to all countries.

    Standard Agent Rule records are scoped to the United States of America (us__v) by default. Edit the Applicable Countries field on an activated rule, or on a copy of a rule, to extend it to additional countries.

  • Applicable User Types - User types the rule applies to. For example, Sales. If left blank, the rule applies to all user types.
  • Display Order - Relative order in which rules are evaluated. For example, if there are four rules with display order values of 1, 2, 7, and 1, they are inserted into the prompt in the order of 1, 1, 2, 7. When two rules have the same display order, their order is determined by which rule is returned first.
  • Status - Rules are enforced when set to Active and disabled when set to Inactive

To make additional changes, for example to the instruction or examples, copy the standard Agent Rule record and edit the copy as a Custom Rule.

Do not activate a standard Adverse Events, Safety or Off-Label, or Promotional Claims rule for the same Applicable Fields as a custom rule used by Agentic Text Monitoring Overview for End Users. Running both agents against the same fields creates duplicate Monitored Text records that require independent triage and escalation. Instead, leave the corresponding standard rule inactive. For more information on Monitored Text records, see Capturing Agentic Text Monitoring Findings.

Custom Rules

Admins create custom rules to determine how records are reviewed for home office monitoring beyond what the standard rules cover. Admins create and maintain their own custom rules.

To create custom rules:

  1. Navigate to Business Admin > Objects > Agent Rules.

  2. Copy an existing rule, or create a new one. The object type must be Agentic Text Monitoring Rule.
  3. Populate the following fields:
    • Rule Name - Name of the custom rule
    • Instruction - Text description of the rule and how the agent should apply the rule. The agent can be instructed to call a prescribing information tool to support monitoring if necessary for the agent's context. To call the tool, include the API name of the tool in the plain language instruction:
      • query_prescribing_information - This tool retrieves PI (Prescribing Information) documents from the vector database for the user's Company Detail products. This tool is useful when the agent needs to perform off-label analysis. See Configuring Agentic Text Monitoring for Home Office for information on uploading PI documents.

      Using the query_prescribing_information tool can increase latency due to retrieval and parsing of PI documents.

    • Acceptable Example - Example of text acceptable under the rule
    • Unacceptable Example - Example of text not acceptable under the rule
    • Applicable Fields - Field and object combinations to be monitored, in a semicolon-delimited list using dot notation. For example, to monitor the Call Text field, enter the text: call2__v.call_notes__v. Text fields on the following objects are supported:
      • Account Plan (account_plan__v)
      • Action Item (action_item__v)
      • Call Report (call2__v)
      • Call Discussion (call2_discussion__v)
      • Call Follow-Up Task (call_followup_task__v)
      • Call Objective (call_objective__v)
      • Commercial Insights (commercial_insights__v)
      • Expense (expense__v)
      • Goal (goal__v)
      • Medical Discussion (medical_discussion__v)
      • Medical Insights (medical_insight__v)
      • SWOT (swot__v)

      Only fields with data type Text, Rich Text, or Long Text are supported for monitoring.

    • Display Order (optional) - Relative order in which rules are evaluated. For example, if there are four rules with display order values of 1, 2, 7, and 1, they are inserted into the prompt in the order of 1, 1, 2, 7. When two rules have the same display order, their order is determined by which rule is returned first.
    • Applicable User Types (optional) - User types the rule applies to. For example, Sales. If left blank, the rule applies to all user types.
    • Applicable Countries (optional) - Countries where the rule applies. This is matched to the User Country. If left blank, the rule applies to all countries.
  4. Ensure the record status is Active.
  5. Select Save.

It may take up to 15 minutes for changes to Agent Rules - including added, deleted, or edited rules - to be reflected in the agent's monitoring results.

Related Topics

Monitoring Submitted Records

Agentic Text Monitoring for Home Office Overview

Flightpath: Agentic Text Monitoring for Home Office