Most AI tasks no longer need a frontier model Read launch post

SIE for Insurance software teams

A claim is a pile of files. The answer still has to point back to the policy.

SIE gives teams building claims and underwriting software one API to read submissions, inspect evidence, retrieve policy language and return cited findings.

Self-host SIE in your infrastructure, or ask about managed deployment through SIE Cloud.

Agent evidence preview
Input
FEMA Appeal B8 + the SFIP

Appeal record, policy clause and two removal estimates

Grounded output
Covered removal stops at the building perimeter

Yard work, barge transport, handling and disposal remain outside the cited scope.

Source attached FEMA appeal decision · policy cited

The operational problem

The model call is only one part of the job.

Claims combine policy wording with photos, estimates, correspondence and prior loss records. The conclusion is useful only when the evidence chain survives review.

01

The file arrives in many formats

Forms, PDFs, photographs, adjuster notes and estimates all describe the same loss differently.

02

Coverage turns on narrow wording

A clause can cover one part of the work while excluding transport, disposal or work outside a boundary.

03

The current policy must control

Endorsements, dates and jurisdiction determine which wording applies.

04

Every finding needs support

Reviewers need the document, passage and item that produced each conclusion.

How SIE helps

Connect the claim file to the governing language

SIE runs document parsing, entity extraction, retrieval and reasoning jobs through one API. Your claims application controls the workflow and review rules.

Your context
Submission file CLAIM
Damage evidence PHOTO
Policy + endorsements POLICY
Prior loss record HISTORY
SIE model jobs · one API
Read forms and PDFs PARSE
Inspect images VISION
Extract scope and amounts ENTITY
Retrieve policy language RERANK
Return cited findings OUTPUT
Your application
A review packet with citations

Claim facts, policy passage, included scope, excluded scope and unresolved questions.

Where it fits

Use SIE inside the product your team is already building.

01

Claim intake

Turn submitted documents into a structured claim record.

02

Policy retrieval

Find the wording that applies to the loss and date.

03

Scope comparison

Compare estimates and evidence with the cited policy language.

04

Adjuster assistance

Prepare a review packet and route uncertain findings.

Representative task benchmark · Redact (PII)

Price, quality and speed for claim documents

One measured model job from this workflow. These figures come from SIE task evaluations, not from the industry example on this page.

SIE GLiNER PII and NuNER Zero Leave 32% less personal data exposed than AWS Comprehend, 79% cheaper

660 synthetic financial documents, 1,792 pieces of personal data. A piece counts when every letter and digit is hidden; punctuation may stay readable.

Price per 1M documents
SIE $73
GPT-6.1 Sol $2,487
Quality Personal data masked
SIE 90.9%
GPT-6.1 Sol 95.7%
p50 latency per query
SIE 61ms
GPT-6.1 Sol —

Recorded example

The agent separates covered removal from excluded transport

The existing example uses a published FEMA flood insurance appeal and the Standard Flood Insurance Policy.

Claim file
Two estimates describe related work

One covers removal under the building. The other adds transport and handling.

Policy
The clause defines a physical boundary

The covered scope runs from beneath the insured building to its perimeter.

Finding
Covered and excluded costs remain separate

The output keeps both the policy passage and the appeal decision attached.

$49,500 removal estimate
$181,832.94 barge estimate
12–15 yd³ stone volume
2 public source documents

Deployment

Run SIE where your data can go.

The workflow uses the same SIE API whether you choose managed infrastructure or run it under your control.

Self-hosted

Run SIE in your infrastructure

Keep models and data inside your environment. Your team controls hardware, networking, access and deployment policy.

View the deployment guide

Contact us

Tell us about your use case and we'll get back to you shortly.

Apply for an inference grant

Free capacity on our hosted cluster for selected projects. Tell us what you run and we reply by email.