Claims & document inconsistency checks

Flag conflicting evidence and missing information across claims and supporting documents for investigation.

Step 01

Define and Align

Review synthetic examples and correct the model’s fields. Explain the evidence and policy distinctions your reviewers use.

Output schema
FindingChoose one

Whether comparable source facts agree, conflict, or require additional evidence.

  • consistent
  • conflicting_evidence
  • missing_evidence
Conflicting fieldsList of values

Fields with an evidenced contradiction about the same fact or event; empty when none.

Missing informationList of values

Documents or source facts required to complete the comparison.

Evidence summaryText

The source context supporting the finding and next review step.

app.sutro.sh/functions/claims-document-inconsistency-checks
Collecting feedback · 1/3
Compare the supplied claim and supporting document under the team’s evidence-review policy.
Example 1 of 3

The same event has two dates

Model input
Review policy
Synthetic document-review policy. Compare only the same field for the same event and case. Different incident dates for the same claim are conflicting evidence. An incident date and a later report issue date can both be correct. If a required supporting document is absent, request it; missing evidence does not establish a contradiction or fraud. Return review findings with the relevant fields and source context.
Primary document
Synthetic claim C-82: incident_date = 2026-09-12.
Supporting document
Repair report for the same claim C-82: incident_date = 2026-09-19. This field explicitly describes the incident, not the report issue date.
Model output
Finding
Conflicting fieldsMultiple values
Missing informationMultiple values
Evidence summary

Step 02

Model Sweep

Compare models on the same expert annotations and choose the quality–cost tradeoff for your workflow.

Compare models on expert agreement and cost. The interactive chart uses illustrative measurements.

Illustrative sweep · select Value or Best to use that model in your function.

Step 03

Deployment

Call Sutro Functions like any other software. Swap the model as soon as something more performant is released.

Selected modelWaiting for sweep results.

Real-time

import sutro as so

input_data = {
    "review_policy": "Synthetic document-review policy. Compare only the same field for the same event and case. Different incident dates for the same claim are conflicting evidence. An incident date and a later report issue date can both be correct. If a required supporting document is absent, request it; missing evidence does not establish a contradiction or fraud. Return review findings with the relevant fields and source context.",
    "primary_document": "Synthetic claim C-82: incident_date = 2026-09-12.",
    "supporting_document": "Repair report for the same claim C-82: incident_date = 2026-09-19. This field explicitly describes the incident, not the report issue date."
}

result = so.run_function(
    "claims-document-inconsistency-checks",
    input_data,
)

Batch

import sutro as so

job_id = so.batch_run_function(
    name="claims-document-inconsistency-checks",
    data="claims_document_inconsistency_checks.csv",
)
results = so.await_job_completion(job_id)

Apply your team’s review policy

Surface contradictory fields and missing documents with their source context. Teach your reviewers’ distinctions between genuine conflicts, different event types, and evidence that has not been supplied. The synthetic examples produce review findings rather than determine a claim outcome.

Ready to get started?

Book time with the team

Join the self-serve waitlist

What are your bottlenecks? (Select any that apply)