Litigation propensity prediction that reads every open injury file each night and raises the ones heading for a lawsuit early
Files turning toward a lawsuit raised early, with the evidence behind each.
See one case, screen by screen ↓It scores each open claim for the chance that it will draw an attorney, a suit or a large verdict, so the file can be worked before that happens. Litigation Propensity Prediction reads every open injury file each night, scores it 0–100 and shows the phrase in the file behind every point.
How to choose AI solutions for insurance →
Updated
The signs are written down. “My lawyer said” in a call note, an MRI and a surgical consult in the medical updates, a voicemail nobody returned, a company truck, a claimant a jury will like, a plaintiff firm with verdicts in the venue. But they sit across notes, letters, bills and records in more than a thousand open injury files, and a supervisor with nine adjusters reviews diaries by sample.
So the turn is usually seen when the summons lands. By then the claimant cannot be contacted directly, the letter of representation is on file, and the cost of resolving the file has already jumped — with defense spend added on top, and a verdict risk that depends on the venue and the firm.
Estimated split for one open bodily injury file, on 30-day diaries by hand and with the nightly read.
Six agents read every open injury file each night and score its risk of suit; the claim supervisor decides.
Claim BI-26-118430: the insured’s 26-ft box truck rear-ended Denise Albright’s Honda CR-V at a red light in Chicago. Liability is clear, there is no letter of representation on file — and in yesterday’s call note, she said a lawyer is advising her. Here is what happens next, screen by screen, in the working solution.
Six agents read every open injury file in the Central casualty unit: 1,284 files, 2,318 new notes, letters and medical updates, 212 signals found. The strip at the top says what moved — 11 files up, 4 down, 24 at 60 or above — and when the next read starts.
Alana starts on the files that moved overnight, not on a diary sample. Denise Albright is first: 41 → 78, up 37. Seven heat cells show where the points came from — lawyer, treatment, delay, claimant, liability, venue, firm — and the note reader replays the reading: each phrase lights in Tom Kessler’s call note and its points tick into the score.
“my lawyer said not to sign anything until after the surgery consult” — Call note · Mon 12 Oct, 16:42
She clicks the lawyer cell. Representation, +19: an attorney is advising Denise but no letter of representation is on file, so direct contact is still allowed under guideline CG-4.3. The phrase is quoted from the note, with the adjuster, the date and the time.
Specials to date $9,860 plus $2,190 wage loss, $1,380 of therapy unpaid since 9 September, BI reserve $28,500, an epidural injection booked for Wed 21 Oct. Each document carries margin notes for the signals it raised, and the risk line runs from 29 at first notice to 78 today.
The MRI report is read with the call note — counted once, inside the +11 for treatment escalating.
The note reader found three phrases in one call note and one voicemail log. The treatment tracker read the MRI of 6 October: L4-L5 herniation, therapy → imaging → injection → surgery raised. The representation watcher found no letter and no records request; the venue and firm profiler named the Circuit Court of Cook County, no firm yet. The risk scorer moved the file 41 → 78 and, past 75, routed it to a specialist the same day.
Next-step writer: “Specialist call today · $38,000–$55,000 now against $120,000–$190,000 if sued · waits for Alana Brooks”
Any signal can be dismissed — read out of context, already handled in a later note, or a duplicate. The score drops by that signal’s points straight away, and the reason is saved to the file for the next weights review.
A specialist calls today and resolves early: $38,000–$55,000, resolved without counsel in 61 of 97 files taken this way since 1 June. Keep the 30-day diary and wait for a demand: $120,000–$190,000 if represented and sued, plus $25,000–$40,000 defense. Or assign defense counsel now, which ends direct contact and signals a fight.
“Why today. Denise says a lawyer is advising her, but no letter is on file, so Calder Ridge can still talk to her directly.”
Alana approves within her $50,000 authority and assigns Priya Natarajan, who is free 10:00–12:00. Priya calls Denise by noon, the two therapy invoices are sent for payment, and the orthopedic plan and MRI images are requested. No release is asked for before the plan is known.
The agents do not call, offer or reserve: Priya calls, Alana approves the offer, Tom updates the reserve.
At-risk files by venue over 12 weeks: Cook County has nearly doubled since July, from 4 files at 60 or above to 7. Brandt & Moreau Injury Lawyers, with 3 verdicts over $1M in the venue, is on 2 of them. Each firm’s record on closed files sits beside it, and the week exports as a report for the large-loss committee.
Tested on 2025 closed files before go-live: 221 of 286 suits — 77% — scored 60 or above at least 30 days before service, with a median warning of 41 days, and 59% of files at 75 or above were sued. Since go-live on 1 June, 168 files crossed 75, 97 were taken by a specialist before a lawyer’s letter, and 61 resolved without counsel.
The same file, seen by the five people who touch it — what their week looked like, and what it looks like now.
Six agents read every open injury file each night and score its risk of suit; the claim supervisor decides.
Reads every new adjuster note, diary, email and call log each night and finds the phrases that signal a lawyer, frustration, delay or escalation — quoting the exact phrase, its document and time.
Finds letters of representation, records requests and preservation letters from firms, lien notices and time-limited demands; on a letter of representation, marks the file represented within 10 minutes and names the firm.
Reads medical bills and records and builds the path of treatment with dates — emergency, therapy, imaging, injections, surgical consult — flagging escalation and gaps.
Finds the venue a suit would be filed in and profiles the plaintiff firm — suits, trials and verdicts — from closed files and published verdict reports.
Adds each signal’s points by family, capped at the family maximum, into a 0–100 litigation risk score with the change since the last read, and routes by threshold: 60 to the heat list, 75 to a specialist the same day.
Finds similar closed files and compares resolving now, waiting and counsel, with cost ranges; recommends one and names who may approve it.
Decides; the manager above $50,000. The agents propose; a named person decides.
The supervisor, adjusters and specialists can ask in plain words. Every answer cites the passage behind it.
Why did Denise Albright’s file jump to 78?
Three things in Tom Kessler’s call note of 12 Oct. “Ortho is talking about injections, maybe surgery”, read with the MRI showing an L4-L5 herniation: +11. Her voicemail of 7 Oct was not returned: +7. And “my lawyer said” — a lawyer is advising her, with no letter on file: +19. At 78 the file goes to a specialist today.
Which Cook County files moved overnight?
Four moved up 10 points or more: Denise Albright 41 → 78, Hannah Liu 49 → 71, Tamika Shaw 66 → 76 and Marcus Tillman 74 → 84 — Tillman now has a letter from Brandt & Moreau. Two moved down: Jamal Whitfield 72 → 62, back to full duty, and Diane Pope 66 → 56 after she agreed to an evaluation meeting.
Can Priya still call Denise directly?
Yes. No letter of representation or attorney notice is on file, so direct contact is allowed. If Denise says she has hired a lawyer, Priya asks for the name and stops discussing the claim, and never discourages her from getting advice.
How early were last year’s suits flagged?
221 of the 286 suits served on 2025 injury files — 77% — scored 60 or above at least 30 days before service, with a median warning of 41 days. Of files at 75 or above, 59% were sued, so the list stays short enough to work.
13 screens from the working solution, on its sample data. Pick one to see it large.
What is read, when a file is raised and who decides are settings. Changes to thresholds and authority need a second approver.
| Setting | Default | Choose from |
|---|---|---|
| Lines in scope | Commercial auto BI · personal auto BI · general liability premises and products | Your lines |
| Read time | Every night at 02:00 | Plus within 10 minutes of a letter of representation |
| Never read | Claims and defense counsel communications | Fixed |
| Heat list | Files at 60 or above | Change, with a second approver |
| Moved overnight | 10 points or more, up or down | Change, with a second approver |
| Same-day specialist | 75 or above (CG-7.2) | Change, with a second approver |
| Who decides | Supervisor up to $50,000 | Claims manager above; claims counsel assigns counsel |
| Never in the score | Race, ethnicity, religion, national origin, sex, income, ZIP code | Fixed |
Every figure is labelled: a target the solution is built to, an estimate, a typical published result, or a proven one.
“demo” = seen in the working solution, on its sample claim files (the 41-day warning is from its test on closed files) · “target” = the design goal, measured in the live solution · “estimated” = our estimate · “typical” = published figures (CLARA Analytics, Litigation product page — represented casualty claims average $77,807 indemnity against $15,936, 390% higher, and run 295% longer; CAS / Triple-I, legal system abuse and liability losses, 2025 — 22.6%–30.8% of commercial auto booked losses). People, companies, claims and law firms named on this page are fictional — characters and sample data in the working solution; counties are real places used as venues only.
It scores each open claim for the chance that it will draw an attorney, a suit or a large verdict, so the file can be worked before that happens. Litigation Propensity Prediction reads every open injury file each night, scores it 0–100 and shows the phrase in the file behind every point.
The agents read the adjuster’s own words and the file’s documents for signals in seven families: representation, treatment escalating, delay and frustration, claimant and injury, liability and insured, venue, and plaintiff firm. Each signal adds points up to its family maximum; “my lawyer said” in a call note, for example, adds 19.
In the working solution: commercial auto bodily injury, personal auto bodily injury, and general liability premises and products. The lines in scope are a setting.
No. Race, ethnicity, religion, national origin, sex, income and ZIP code are never read or used, and that setting is fixed. The claimant facts that count are the ones a jury sees — occupation, injury and dependants — and they are listed on the file.
No. The agents never contact a claimant, make an offer, change a reserve or assign counsel. The next-step writer drafts options with cost ranges; the claim supervisor decides within their authority, the claims manager above it, and claims counsel assigns counsel.
The Results screen shows it. In the working solution, a test on 2025 closed files found 221 of 286 suits scored 60 or above at least 30 days before service, with a median warning of 41 days. Signal weights change only after a test on closed files and a second approver.
Yes. The heat list threshold, what counts as moved overnight, the score that sends a file to a specialist the same day, and who approves at which amount are settings. Changes need a second approver.
The Agentic Solution Engine builds and deploys it from your requirements — your claims guidelines, authority table and a sample of closed injury files — and it goes live once every quality gate has passed. We will run it on your own closed files first.
Thousand-page demands read in full and answered before the deadline.
1-min demoTime-Limited Demand ManagementNo policy-limits deadline missed: every demand on a clock, every condition checked.
1-min demoBodily Injury Claim ValuationSettlement authority set from comparable closed files and local verdicts.
1-min demoWe’ll run Litigation Propensity Prediction on a sample of your own closed injury files.