Complete AI Training
Sign inGet my AI kit

Your job's AI kit

Get your AI kit

Tell us who you are and what you do. We show you your kit right away and email you the link: skills, prompts, AI agents, MCP servers and courses for your job.

500+ jobs ready, and we make a kit for any other job. No payment needed to look.

Share

AI agent for underwriters

Loss Run Analysis Agent

A clean, checked loss history with frequency, severity and large losses

Data cleanupAnalysisInsurance
See how it worksGet access
Loss Run Analysis Agent: what goes in, what the agent does and what you get

What it does

Loss runs arrive from prior carriers as PDFs in all kinds of layouts, and normalizing them takes hours. This agent reads each loss run and puts claim dates, causes, status and amounts into one standard table. It then checks that its totals tie to the summary figures printed on the loss run. If they do not, it flags the page and reads it again, using a second method for tables that were hard to read. It computes frequency, severity and large losses and writes a short analysis. The underwriter approves the analysis used in the decision. Edge case: a claim appears on two pages with different amounts, so the agent keeps the later one and notes the conflict.

How it works

Follow the arrows from top to bottom. The orange dashed arrow is the loop: when a check fails, the agent goes back and tries again.

Start and resultWhat it doesA check on its own workWaits for your OKGoes back and retries
Yes, continueYes, continueApprovedNoNo 1 STARTS WHEN Loss runs received 2 USES A TOOL Read each loss run page and extract claims 3 DOES Normalize dates, causes, status and amounts into onetable 4 USES A TOOL Read the carrier's summary totals 5 CHECKS THE RESULT Do the extracted totals tie to the printed summary? If not: flag the page that differs and read it againwith a second method. Back to step 2. 6 DOES Remove duplicate claims and note conflicts 7 DOES Compute frequency, severity and large losses by year 8 CHECKS THE RESULT Do yearly counts match the policy periods covered? If not: fix the claim dating and recompute. Back to step6. 9 YOU APPROVE Underwriter approves the analysis 10 RESULT Loss analysis added to the underwriting file
Read the steps as a list
  1. Loss runs received
  2. Read each loss run page and extract claims
  3. Normalize dates, causes, status and amounts into one table
  4. Read the carrier's summary totals
  5. Do the extracted totals tie to the printed summary?If not: flag the page that differs and read it again with a second method. Back to step 2.
  6. Remove duplicate claims and note conflicts
  7. Compute frequency, severity and large losses by year
  8. Do yearly counts match the policy periods covered?If not: fix the claim dating and recompute. Back to step 6.
  9. Underwriter approves the analysisThe agent waits here for your OK.
  10. Loss analysis added to the underwriting file

How it decides

Totals must tie to the carrier's summary within a small tolerance. Claims above the large loss threshold are listed separately.

  • Treat totals as tied when they match within one percent
  • Keep the later amount when duplicate claims differ
  • List losses above the threshold separately
  • Flag any gap in years covered

Make it yours

Every agent is a starting point. You choose these settings for your own situation.

  • Large loss threshold (default: $50,000)
  • Tie-out tolerance (default: 1%)
  • Years to analyze (default: five)
  • Cause categories

What keeps you in control

It always asks you first

  • Underwriter approves the analysis before it is used

Hard limits

  • Never changes the carrier's data
  • Never decides coverage or price

It stops when

  • Done: totals tie and the analysis is approved
  • Stop: loss run pages are unreadable

Set it up

We guide you through the set-up, step by step

Members get the full set-up guide for this agent. No technical skills needed: you copy, paste and upload.

10 minto set it up in your AI
5 AIsChatGPT, Claude, Copilot, Gemini, Grok
  • One set of instructions to paste into your AI, with the clicks for ChatGPT, Claude, Microsoft 365 Copilot, Gemini and Grok
  • The agent then walks you through connecting your own data, one source at a time
  • A downloadable copy with the flow chart, the rules and the full guide
Get access to this agent

An example run

What happensA five year loss run listed 23 claims totaling $412,600. The agent extracted $388,100, so the tie-out failed. It re-read page three and found a $24,500 water claim cut off at the page edge. Totals then matched. It found two losses above $50,000, and frequency of 4.6 claims a year. The underwriter approved the analysis.

More agents for underwriters

Quality control · Customer support

Policy Change Impact Check Agent

Show the customer the real effect of a policy change before it is bound.

10 steps2 checks4 jobs
Risk management · Insurance

Application Versus External Data Agent

A short list of confirmed differences between the application and outside records, with evidence

9 steps1 check2 jobs
Quality control · Compliance

Policy Wording and Endorsement Check Agent

A quote whose wording matches approved forms, with agreed exclusions present and exceptions approved

10 steps2 checks1 job
Compliance · Writing

Declination and Reconsideration Agent

A documented decline and a fair, recorded recheck of any new information

11 steps2 checks1 job