Skip to content

Claims First-Notification-of-Loss Intake

Example prompt: "When a client emails our claims address or submits the claims web form, draft the first-notification-of-loss email to the insurer, send the client an acknowledgement explaining what happens next, log the claim to our register, and keep it in our weekly client status digest until it closes."

The Problem

A client calls at 4pm on Friday to report a burst pipe at one of their offices. The loss adjuster needs an FNOL email with the policy number, the date and time, the location, what happened, what the immediate exposure is, who is in charge of mitigation, and the photos — and the insurer's contact details are in last year's binder. The account handler who knows the file is in a meeting. The client wants to know when someone will get back to them and whether they should be moving the staff to the back of the building. The cases that bite are the ones where the FNOL did not reach the insurer until Monday morning and the mitigation invoice the client paid out of pocket on Friday night becomes a coverage argument because the insurer was not put on notice in time.

How GloriaMundo Solves It

We build a workflow that fires the moment a claims email lands in the dedicated claims mailbox or the client submits the claims web form. An integration step reads the email or the form fields. A code step looks up the client, the policy, the insurer's claims contact, and the broker reference from the Policies sheet. An LLM step composes the FNOL email body with every field the insurer needs in the first 24 hours, attaches the photos from the form's Drive folder, and saves it as a Gmail draft to the insurer. A second LLM step composes a short acknowledgement to the client confirming we have notified the insurer (as a draft for the broker to send) and explaining what the next 48 hours will look like. An integration step logs the claim to the Claims Register with a provisional claim_ref. Glass Box preview shows the FNOL draft, the client acknowledgement draft, and the register entry before any of them leave the office.

Example Workflow Steps

  1. Trigger (integration): A new email to the claims@ mailbox or a submission to the claims Google Form.
  2. Step 1 (integration): Read the email body and attachments, or the form fields — client_name, policy_number_or_known_reference, date_of_loss, location, description, immediate_mitigation_taken, photos, contact_name, contact_phone, contact_email, urgency_self_rating.
  3. Step 2 (integration + code): Look up the policy on the Policies sheet, the insurer's claims contact and broker reference on the Insurers sheet, and the client's account handler on the Clients sheet. If the policy number is not recognised, route to the account handler for manual policy match instead of drafting.
  4. Step 3 (code, gate): Dedupe on gmail_message_id (or form_response_id) against the 'Claims Register' tab — a final state ('Notified' or later) halts; a provisional state ('Allocating') resumes; no row writes a provisional row with status 'Allocating' before any subsequent step and allocates claim_ref as 'CLM-[YYYY]-[6-digit sequence]'.
  5. Step 4 (llm): Compose the FNOL email body for the insurer — policy number, broker reference, client name, date and time of loss, location, description in the client's words, immediate mitigation taken, photos attached, drivability or operations impact, contact details for the client, and a request for the insurer's claim reference and loss adjuster appointment.
  6. Step 5 (integration): Save the FNOL as a Gmail draft to the insurer's claims address with photos attached from the form's Drive folder. Persist gmail_fnol_draft_link to the provisional Claims Register row.
  7. Step 6 (llm): Compose the client acknowledgement — confirms we have prepared the insurer notification, summarises what the insurer will typically ask for next, names the account handler and gives a direct contact, and (where the urgency self-rating is high) confirms the broker is monitoring through the weekend.
  8. Step 7 (integration): Save the client acknowledgement as a Gmail draft to the contact_email. Persist gmail_client_draft_link to the provisional Claims Register row.
  9. Step 8 (conditional + integration): If urgency_self_rating is 'high' or the description mentions injury or business interruption, post a higher-priority page in #claims-urgent on Slack tagging the account handler with the claim_ref, the location, and links to both drafts.
  10. Step 9 (integration): Flip the provisional Claims Register row's status from 'Allocating' to 'Drafted'. The insurer's claim reference gets written back to the row by the account handler (or a separate listener on the insurer's reply email) once the claim is opened.

Integrations Used

  • Gmail — the claims mailbox trigger, the FNOL draft to the insurer, the acknowledgement draft to the client
  • Google Forms — the alternative claims web form (optional)
  • Google Sheets — the Policies, Insurers, Clients, and Claims Register
  • Google Drive — the form's photo upload folder and the per-claim document folder
  • Slack — the urgent page to the account handler for high-severity cases

Who This Is For

Commercial brokers, MGAs, and in-house risk managers handling a steady flow of property, liability, motor fleet, or business-interruption claims from SME and mid-market clients, where the difference between an FNOL going out within an hour and one going out the next morning is the difference between a clean claim and an avoidable coverage argument.

Time & Cost Saved

A claims handler pulling an FNOL together from a client email, the Policies sheet, and last year's broker binder typically loses 30 to 45 minutes per claim, and the client acknowledgement that should have gone out within an hour is often the first thing that slips. This workflow puts the FNOL draft, the client acknowledgement, and the register entry in front of the handler within minutes of the claim landing, and the urgent cases page the account handler before the client picks the phone up again. The per-claim admin overhead drops from 45 minutes to under 10, and the cases that previously got delayed because the handler was on another call no longer wait.