How to build a FNOL Report Intake and Triage Agent
This workflow streamlines the First Notice of Loss (FNOL) process by capturing report details, extracting key information, triaging based on urgency, and presenting a clear summary with suggested next steps.
Challenge
Ensuring FNOL report details are complete and accurate is difficult, as users often provide vague or missing information at this first step, before a claim is officially created. This makes reliable extraction and triage essential to avoid delays or misclassification of high-priority reports.
Industry
Insurance
Department
Legal
Integrations
OpenAI
Google Docs
Gmail
Workflow Overview
1. Trigger: Email Intake
On Email Received
The workflow starts automatically whenever an email is received in the connected Gmail inbox (specifically, for FNOL reports).
2. Claim Extraction & Triage
OpenAI LLM (llm-0): Extract and Triage Claim
Extracts all relevant claim details from the email (claimant name, contact, incident date, loss type, description, location, urgency, etc.).
If information is missing, it generates follow-up questions.
Enriches the claim by querying Salesforce for additional claimant/claim info.
Consults a knowledge base (using urgency criteria) to help categorize the claim’s urgency.
Outputs a structured summary with urgency and recommended actions.
3. AI Routing: Urgency Classification
AI Routing Node
Receives the structured claim summary.
Classifies the claim into one of four urgency categories:
High Urgency
Medium Urgency
Low Urgency
Critical
4. Branching Actions Based on Urgency
A. High Urgency
Create Google Doc (action-2):
Generates a Google Doc for internal record-keeping with all extracted information.
High Urgency Email (llm-4):
Drafts a customer-facing email, confirming receipt and explaining that the claim is high priority.
Send Email (action-7):
Sends the high-urgency response to the policyholder.
B. Medium Urgency
Create Google Doc (action-1):
Generates a Google Doc for internal record-keeping.
Medium Urgency Email (llm-2):
Drafts a customer-facing email, confirming receipt and explaining that the claim is medium priority.
Send Email (action-4):
Sends the medium-urgency response to the policyholder.
C. Low Urgency
Create Google Doc (action-3):
Generates a Google Doc for internal record-keeping.
Low Urgency Email (llm-3):
Drafts a customer-facing email, confirming receipt and explaining that the claim is low priority.
Send Email (action-5):
Sends the low-urgency response to the policyholder.
D. Critical
Critical Response (llm-1):
Drafts a message instructing the user to call emergency services if in danger.
Send Email (action-6):
Sends the critical response to the policyholder.
5. Output Node
Output Node (out-0):
Presents a formatted summary of the claim triage for review or logging.
Key Integrations & Automations
Salesforce: Used to enrich claim data.
Google Docs: Automatically creates documentation for each claim, categorized by urgency.
Gmail: Sends personalized, urgency-specific responses to policyholders.
Knowledge Base: Used to define urgency criteria for accurate triage.
Summary Table
Step | Action |
|---|---|
1. Trigger | Email received in Gmail (FNOL report) |
2. Extraction | LLM extracts claim details, enriches with Salesforce, checks urgency via knowledge base |
3. Routing | AI classifies claim as High, Medium, Low, or Critical urgency |
4. Branching | For each urgency: creates Google Doc, drafts & sends tailored email to policyholder |
5. Output | Outputs formatted triage summary |
In short:
This workflow automates the intake, triage, documentation, and customer communication for insurance FNOL reports, ensuring every claim is processed efficiently and every policyholder receives a timely, personalized response based on the urgency of their situation.
