Adverse Event Management Agent

The Adverse Event Management Agent evaluates the Interaction for a potential adverse event (AE), including the subtle, indirectly worded cases a specialist might overlook, and recommends an adverse event for review. It grounds that assessment in the regulatory definition of an adverse event and the Interaction's own notes.

How the agent works

Following its reasoning instructions, the agent runs three actions in order:

1. Retrieve the Interaction with the AIMI Agent: Get Interaction Details flow.
2. Pass those details to the Interaction: Adverse Event Recommendation prompt template, which assesses whether an adverse event may have occurred and, when it recommends one, drafts a de-identified clinical narrative.
3. Call the AIMI Agent: Create AE Response Log flow to write the prompt's response to a new AIMI Response Log of record type MAI_Adverse_Event.

The agent recommends at most one adverse event per Interaction. The response log captures the agent's recommendation verdict (MAI_Recommendation__c), its justification (MAI_Justification__c), the de-identified clinical case narrative (MAI_Narrative__c, clinical substance kept, identifying detail removed for a safety reviewer to restore downstream), and the suspect product (MAI_Product__c) when one is resolvable.

In the review panel

The AE recommendation appears as a card in the AIMI Recommends panel. The specialist can:

• Accept and Create, when the recommendation is Recommended, a MED_Adverse_Event__c record is created; when a product also resolves, a related MED_AE_Drug_Information__c record is created for its matching MED_Product__c record. The product is matched by name or trade name for the Interaction's country (MED_Country__c), falling back to a country-agnostic base product when no country-specific match exists.
• Reject, decline the recommendation; a rejection rationale is required and is stored in MAI_Rejected_Rationale__c.
• Check Source, open the originating email to confirm what the agent based its assessment on.

This agent is also available on demand through Ask AIMI (Detect Adverse Event).

Actions available for reasoning

Action	API name	Type
AIMI Agent: Get Interaction Details	MAI_AIMI_Agent_Get_Interaction_Details	Flow
Interaction: Adverse Event Recommendation	MAI_Interaction_Adverse_Event_Recommendation	Prompt template
AIMI Agent: Create AE Response Log	MAI_AIMI_Agent_Create_AE_Response_Log	Flow

warning

The agent's AE recommendations are a drafting aid, the filed AE record remains the regulated source of truth.

See Recommendation types → Adverse Event for the full field reference.