Introducing Conversational First Notice of Loss (CFNOL)

Hi Marley’s Conversational First Notice of Loss (CFNOL) streamlines claims starting at FNOL via a simple, conversational solution with an empathetic touch.

Customers explain what happened in their own words and carriers gather critical information immediately. Any operator assigned to the Hi Marley case can access and reference the customer stories and data throughout the claim lifecycle.

Core CFNOL Steps:

Trigger
Story & Photo Analysis
Follow-up questions
End of workflow actions

1 - Trigger

There are two ways for the policyholder to trigger the Conversational FNOL workflow to start:

Phone Call - a policyholder who calls the carrier can make a selection in the carrier's IVR system that triggers a corresponding CFNOL workflow to begin (ex: "Push 1 to text your FNOL...")
Inbound Keyword - an end user can send an inbound keyword via text which triggers a corresponding CFNOL workflow to begin (ex: policyholder texts "AUTO" to 555-555-5555 report an auto claim)

2 - Story & Photo Analysis

Story Capture

The first source of FNOL information Hi Marley captures is the policyholder's story.

Phone Call - Hi Marley captures the story over voice if the policyholder initiated CFNOL with a phone call. Hi Marley transcribes the story and store both the audio recording and transcription on the associated case.
Inbound Keyword - Hi Marley captures the story over text if the policyholder initiated the CFNOL with an inbound text message

Photo Capture

After story capture, the policyholder is asked to submit any photos of the incident they may have. The policyholder can upload as many photos (or videos) as they want.

AI Analysis of Story & Photos

FNOL data is defined by a carrier at the time of workflow creation and represents all data the carrier would like to obtain during the CFNOL.

Hi Marley attempts to identify as much of the FNOL data as possible from the policyholder's story and photos as noted below:

Story analysis - In all FNOL workflows Hi Marley uses AI to analyze the policyholder's story for any relevant FNOL data. All data identified via story analysis is stored and visually denoted to show it was identified by AI.
Photo analysis - In Auto FNOL workflows Hi Marley use AI to analyze photos for FNOL data. (Property workflows do not currently support AI photo analysis). All data identified via photo analysis is stored and visually denoted to show it was identified by AI.

AI Analysis Summary

The policyholder receives a summary of all the data identified from their story and photos. They are able to respond with any corrections. Based on the policyholder's response to the summary Hi Marley attempts to update any relevant FNOL data values.

3 - Follow-up questions

Gathering remaining FNOL data

Hi Marley will attempt to gather FNOL data not gathered in the story and photo analysis by asking the policyholder a series of follow-up questions. When the policyholder responds to a question with a valid response (see validation section below) they will be sent the next question.

The questions are predetermined which, means the questions are NOT generated by AI.

CNFOL - Gathering remaining FNOL data.png

Follow-up question summary

After all questions have been answered a final summary is sent to the policyholder containing all data provided during follow-up questions. Based on the policyholder's response to the summary Hi Marley attempts to update any relevant FNOL data values.

4 - End of workflow actions

When the workflow ends the following actions are triggered:

Email summary (optional) - If the CFNOL workflow was successfully completed an email summarizing the FNOL data that was gathered will be sent to defined email address(es).
Webhooks - A set of webhooks surfacing the FNOL data will be triggered any time the workflow ends, regardless of if it it was successfully completed.

Hi Marley's Conversational FNOL uses Responsible AI

Conversational FNOL uses what we’ve termed as Responsible AI:

We don't send AI-generated text or speech directly to a policyholder, so there’s no chance of the AI hallucinating something to the customer. All messages sent to the policyholder are pre-determined.
We’re using AI to ingest, interpret, and structure policyholder speech and text.
We play back all information we collected from the policyholder in a summary texts to the policyholder, and ask the policyholder to confirm that the information is correct, and if not, to make any corrections.