Insurance and in-house legal
Your insurer’s own AI.
Reads claims and contracts, flags risk and finds the past case that matches, in your own wording.
Three deviations: a liability cap above your standard 1, 30 days’ notice instead of 90 2, and no audit right 3.
- 1Contract playbook, suppliers
- 2Notice clause, standard
- 3Audit clause, standard
Illustrative
- Tasks to start with. A Test Lab takes one.
- 3
- The Formula it starts withWith JLM
- RLM
- To roll back to the last release, by design
- 60 s
- Eigna Hybrid. Data stays in the region you choose.
- Hybrid
The work that cannot leave
Claims, policies and contracts hold your customers’ private lives and your own pricing. They should train only your model.
- Claims
- Policies and wordings
- Contracts
- Customer correspondence
Where it runs, for insurance
Eigna Hybrid
Data stays in the region you choose.
Small experts on your machines; the main expert on a GPU in your region.
Prompts for the main expert travel over mutual TLS, with no prompt logs by default.
The first expert
Three tasks to start with.
Decisions and reviews approved by your claims handlers and in-house lawyers.
Contract review
Reads against your playbook and flags what deviates.
Three deviations: a liability cap above your standard 1, 30 days’ notice instead of 90 2, and no audit right 3.
- 1Contract playbook, suppliers
- 2Notice clause, standard
- 3Audit clause, standard
Learns from · Reviews your lawyers approved.
Claims
Sorts and checks claims, and finds the past case that matches.
Covered under the building policy, less the excess 1. A similar claim was settled after a plumber’s report 2.
- 1Building policy, section 3
- 2Closed claim, 2024
Learns from · Claims your handlers decided.
Policy questions
Answers from your own policy wordings, with the clause it rests on.
Yes, if the home was empty for fewer than 60 days 1. A longer absence needs the heating check in clause 7 2.
- 1Home policy, section 4
- 2Clause 7, unoccupied homes
Learns from · Answers your claims team approved.
The proof we will measure
A metric, a baseline and a sealed test, agreed before we start.
The target is set with you before the Test Lab starts. If the expert does not beat your baseline, you keep the report.
17.8%
Our first training run cut forecast error 17.8% against its base model. It is not yet ahead of a simple last value baseline, and we publish that too.
Rules in plain words
What the rules ask, and what every release gives you.
- 01
Privilege stays with you
Your files train only your model. Client data never moves between clients.
- 02
Every answer names its source
The clause, the memo or the claim it came from, so a lawyer can check it in seconds.
- 03
A record of every release
A named owner, a release record and rollback in 60 seconds. Designed to support your compliance work.
Insurance and in-house legal
Your files train only your model.
Claims, policies and contracts never move between clients, and every answer names the clause, the memo or the claim it came from.
Insurance: questions and answers
Does our work train anyone else’s model?
No. Your files train only your model, and only the examples your people approve.
Can we keep the most sensitive files on site?
Yes. With Eigna Hybrid, small experts run on your machines and the main expert in the region you choose. With Eigna Local, everything stays on your machines.
How is review quality measured?
On a sealed set of your past files: reviews approved without rework and time per file, against how your team works today.
Test Lab
Request a Test Lab for insurance.
One task, your data, a sealed setup, about four weeks. If the expert beats your baseline, the test set becomes your first release gate. If not, you keep the report.
- RouteDirect, and through insurance and legal tech partners.
- Watch forFast to prove, because the work is document-heavy and measured file by file.
- ReplyWithin two working days, from a founder.
