Claims Reserving with AI: Where It Helps and Where It's Dangerous

Claims reserving is where an insurer decides how much money to set aside for claims it knows about but hasn't fully paid, and for claims that have happened but haven't been reported yet. It's one of the most consequential numbers a carrier produces — it drives solvency, pricing, and how profitable the business even looks.

AI is arriving in reserving, and the pitch is seductive: more accurate reserves, faster, updated continuously. Some of that is real and valuable. Some of it is genuinely dangerous. Knowing the difference matters more here than almost anywhere else in insurance AI, because the failure mode isn't an annoyed customer — it's a mis-stated balance sheet.

Where AI genuinely helps reserving

Individual claim reserving. Traditional reserving works on aggregates — triangles of how claim cohorts develop over time. But two claims in the same cohort can be wildly different: a minor fender-bender and a claim that's quietly heading toward litigation. AI can read the specifics of an individual claim — its characteristics, its narrative, its early signals — and set a more accurate case reserve than a one-size-fits-the-cohort figure. This is real, useful, and lower-risk, because a human adjuster reviews it.

Early warning on development. Models can flag claims likely to deteriorate — the ones that look small now but have the hallmarks of becoming large. Surfacing those early improves both reserving and claims handling. Also genuinely valuable.

Faster IBNR signals. "Incurred but not reported" — claims that have happened but you don't know about yet — is estimated from patterns. AI can incorporate more signals and update more frequently than quarterly triangle exercises, giving earlier warning of a shifting trend.

These share a shape: AI informs the reserving process with better granularity and earlier signals, while actuaries and their methods remain in control of the number.

Where it gets dangerous

The danger begins when AI moves from informing the reserve to setting it, autonomously, at the aggregate level that hits your financial statements.

1. Reserves are a regulated, audited number. They're not an operational metric you can tune and observe. They determine solvency and get scrutinised by regulators, auditors, and rating agencies. A number that can't be explained and defended isn't acceptable, however accurate it might be. "The model set it" is not a position you can take to an auditor about your balance sheet.

2. Black-box reserves are unauditable. If your reserve is the output of a model nobody can fully explain, you cannot answer the questions an auditor will ask about it. Reserving demands explainability not as a nicety but as a condition of sign-off.

3. Models trained on the past miss regime changes. Reserving's biggest risk is a shift the historical data doesn't contain — a new litigation trend, a change in claims inflation, a legal ruling that expands liability. An AI extrapolating from history will confidently under-reserve into exactly the environment that hurts you most, precisely because it's never seen it. Actuaries apply judgment about the future; a model applies patterns from the past.

4. The feedback loop is brutally slow. If a reserving model is wrong, you may not find out for years, as claims develop. By then the mis-statement has propagated into pricing and capital decisions. You cannot iterate your way to a good reserving model the way you can with, say, a fraud model that gives feedback in weeks.

The pattern that keeps AI safe here

The right architecture mirrors the governed pattern used everywhere in regulated insurance AI: the model informs; the regulated number stays under human control.

  • AI sets granular case reserves that adjusters review — operational, reviewable, valuable.
  • AI provides signals and early warnings into the actuarial process.
  • Actuaries, using established and defensible methods, set the aggregate reserves that hit the financials — informed by the AI, not replaced by it.
  • Every AI contribution is explainable and logged, so the final number can be defended.

This isn't AI-timidity. It's recognising that the reserving number has a different risk profile from a claims-triage decision, and the level of autonomy should match the stakes and the reversibility.

And, underneath it all

Even the safe, valuable applications depend on the same thing everything else does: clean, granular, well-structured claims history. Individual-claim reserving needs individual-claim data — including the narratives and characteristics usually trapped in free text and documents. Feed a reserving model the fragmented, dirty history most insurers actually have, and its granular reserves are granular fiction.

So the sequence, as ever, is foundation first: get the claims data clean and structured, apply AI where it informs and a human reviews, and keep the regulated aggregate number under actuarial control. Do that and reserving gets genuinely better. Skip it — let a black box set the number on dirty data — and you've automated your way to a balance-sheet risk you won't detect until it's expensive.

Reserving is the clearest example of a rule that runs through all of insurance AI: match the AI's autonomy to the cost of being wrong. Here, the cost of being wrong is your solvency. Govern accordingly.

We build the clean, granular claims data foundations that reserving AI depends on. More at IntelliBooks.

Comments

Post a Comment

Popular posts from this blog

Why Insurance Fraud Rings Are a Graph Problem (And Your SQL Database Can't See Them)

Your fraud team is good at catching individuals. A claimant whose story shifts, a receipt that looks altered, someone who files a bit too often. That's what the rules were built for, and they work. They're nearly blind to organised fraud — and organised fraud is where the money goes. The reason isn't that your team lacks skill or your rules lack sophistication. It's that you're asking a question your database structurally cannot answer. What a fraud ring actually looks like A ring isn't one bad claimant. It's a network: a handful of claimants, a couple of "witnesses", a repair shop, a medical clinic, a few bank accounts, some phone numbers, and a device or two — recombined across dozens of claims over months. Take any single claim from that ring and it looks fine. The claimant has no history. The damage is plausible. The repair estimate is in range. Nothing trips a rule, because the rule is examining one claim in isolation, and in isolati...
Read more

Claims Leakage: The Quiet Loss AI Can Close

Claims leakage is the money your company pays out that it didn't have to. Not fraud — leakage. The extra amount on legitimate claims that got overpaid because someone missed something, applied the wrong rule, or lacked the information to know better. Industry estimates put it at 3–10% of claims spend. For a mid-size insurer that's a rounding error you could retire. And almost nobody manages it directly, because it's nearly invisible — every individual leaked rupee looks like a normal payment. Why leakage is so hard to see Fraud has a villain. Leakage doesn't. It's the accumulation of small, defensible, individually-reasonable decisions: An adjuster settled at the reserve figure because they were busy, and the reserve was set high. A repair was authorised at shop rates when a network rate applied. A subrogation opportunity was never flagged, so recovery never happened. A policy limit or deductible was applied wrongly because the coverage detail was buried...
Read more

The Subrogation Money Insurers Leave on the Table

Subrogation is the least glamorous word in insurance and one of the most profitable. It's the process of recovering money from the party actually responsible for a loss, after you've paid your customer. Done well, it flows straight to the bottom line — recovered money is nearly pure margin. Most insurers leave a lot of it on the table. Not because they don't try, but because the opportunities are identified too late, by the wrong people, from incomplete information. It's a data problem, and it's an unusually clear one. Why subrogation is a timing game you're losing Recovery potential decays fast. The evidence that establishes another party's liability — the scene, the witnesses, the vehicle, the maintenance records, the CCTV — degrades or disappears within days. The other party's insurer gets less cooperative once they've closed their own file. So the single biggest lever in subrogation is identifying the opportunity early , ideally at first ...
Read more