Carriers · state funds · third-party administrators · MGAs
LAE runs 10–15% of losses paid. On claim files and medical records, automation saves US$300–500 per claim in labour against roughly US$10 per claim in technology. A 30-to-50x ratio needs no modelling assumptions to be persuasive — it needs your volume and your fully loaded cost per file.
Why we do not lead with catastrophe
Every vendor in this market waits for a season and then sends the same email about claim surges. You can see it coming, and it makes the sender look opportunistic rather than useful.
Loss adjustment expense is true in every season. It is on the face of your combined ratio, it is measured the same way at every carrier, and it moves whether or not a storm makes landfall. When a catastrophe does arrive, straight-through processing is the only lever that scales inside the event window — but that capability has to be built before the water rises, not procured while it is rising.
If we only become interesting to you in September, we were never interesting.
The hard content
It is handwritten by a physician in a hurry, photocopied twice, faxed once, annotated in the margin, and it determines whether an indemnity payment is correct. It arrives in a stack of two hundred pages of which nine matter.
The reason this has resisted automation is not that machines cannot read handwriting — they can now, cheaply. It is that being wrong is expensive and invisible. A misread date of service or a missed causation statement does not announce itself; it surfaces eighteen months later in a reserve development or a bad-faith allegation.
So the product is not the reading. It is the confidence score on every extracted field, the routing of anything uncertain to a named adjuster with the source page attached, the immutable record of what the machine decided and what the human changed, and the correction feeding back so the same mistake is not made on the next file.
A life insurer's claims document estate — medical evidence, policy documentation and beneficiary records — processed end to end. The closest available analogue to a US life underwriting evidence pack or a workers-comp medical file.
Your fully loaded cost per claim file today, and the percentage of files that get reopened. Those two figures decide whether this is worth either of our time, and you already know both.
Third-party administrators and claims outsourcers
A carrier that automates claim handling books the saving itself. A TPA that automates claim handling hands the saving to its client at renewal — which is why an efficiency pitch to an administrator is, read carefully, a request to lower your own price.
The relevant framing is different. Well-funded AI companies are now positioning openly against the outsourcing industry and telling your clients that the work can be done without you. Whatever those claims are worth today, your clients have heard them, and the renewal conversation has already changed.
Densery is capacity you can sell: the same adjusters handling materially more files, with an evidence record you can put in front of a carrier's audit team. One deployment covers every client at once, because the document types repeat.
Fit
| Organisation | Size band | Why this band |
|---|---|---|
| P&C, life and workers-comp carriers | Above US$250M in premium | The claim and underwriting file volume needed to clear the technology cost several times over. |
| State funds and state entities | Any scale | Data residency is mandatory rather than preferred, procurement is slow but genuine, and volume is exceptional. |
| TPAs, MGAs and claims services | Above US$100M in revenue | Multi-client document operations where one capability serves many carrier relationships. |
| Not a fit | Top-5 national carriers; digital-native insurtechs | The largest carriers run internal data-science organisations under multi-year agreements — we reach them through their TPA and claims-vendor relationships instead. Digital-native insurers have structured data and no document estate to fix. |
The next step
A scoping session is not a demo. Bring twenty real files, redacted if you need to. We take three numbers off you — annual volume, fully loaded cost per file today, and what happens when the output is wrong — and hand back a one-page value case in your own KPIs.
If the arithmetic says we are not a fit, we will tell you in the room rather than six weeks later.
If your data can go anywhere and your documents are already clean and digital, you do not need us. Use a hyperscaler document API and spend the money on something harder.
Densery