Audit a Quarter's Insurance Claims

Batch-review insurance claims for compliance and fraud indicators.

Who is this for

Claims Auditor / Compliance Manager at an insurer. You need to review a quarter's worth of claims to check for compliance with ICOBS regulations and identify potential fraud patterns before they escalate.

What you'll do

Batch-submit claims files for deep audit with fraud detection
Receive compliance findings and fraud indicators per claim
Receive full report data as part of each check response — no separate step needed

Submit claims for audit

Loop through your claims files, submitting each for a deep audit. Tag every job with the audit period so you can filter results later. Each check response includes full report data — no separate generate step is needed.

audit_claims.py

python

from curvestone import Agent

agent = Agent()

# Submit claims for audit
claims = [
    {"ref": "CLM-2026-001", "docs": ["claim_form_001.pdf", "assessor_report_001.pdf"]},
    {"ref": "CLM-2026-002", "docs": ["claim_form_002.pdf", "police_report_002.pdf"]},
    # ... up to hundreds of claims
]

results = []
for claim in claims:
    result = agent.check(
        case_type="motor_claim",
        depth="deep_audit",
        documents=[open(d, "rb") for d in claim["docs"]],
        reference=claim["ref"],
        metadata={"audit_period": "Q1-2026"},
    )
    results.append(result)

# Each result already includes full report data
for r in results:
    print(f"Job {r.id}: triage={r.triage}")

Claim audit result

Each claim returns compliance findings and fraud indicators. Claims flagged for fraud review include a flags array in the response.

claim_result.json

json

1{
2  "id": "job_5kRx2mNpQ8",
3  "type": "check",
4  "status": "completed",
5  "triage": "red",
6  "reference": "CLM-2026-002",
7  "processing_time": "178s",
8  "checks": [
9    {
10      "name": "Claims Handling Compliance",
11      "triage": "green"
12    },
13    {
14      "name": "Reserve Adequacy",
15      "triage": "amber",
16      "finding": "Initial reserve £2,400 below assessor estimate"
17    },
18    {
19      "name": "Fraud Indicators",
20      "triage": "red",
21      "finding": "Inconsistent timeline: police report filed 3 days before incident date on claim form"
22    },
23    {
24      "name": "Subrogation Opportunity",
25      "triage": "green"
26    }
27  ],
28  "flags": ["fraud_review_recommended"],
29  "cost": "£8.00"
30}

Batch results

Each check response includes full report data. Aggregate the results across your batch to build quarterly summaries — no separate generate step is needed.

What happens

Deep audit suite

The deep_audit depth runs the full suite of insurance compliance checks including claims handling compliance, reserve adequacy, fraud detection, and subrogation opportunity analysis. This is the most thorough level of review available.

Fraud detection

Claims flagged for fraud review are marked in the response with a flags array containing fraud_review_recommended. The agent cross-references dates, amounts, and narratives across all documents in the claim to identify inconsistencies that warrant human investigation.

Report data

Each check response includes full report data with triage outcomes, fraud flags, and key findings. Aggregate batch results to build quarterly summaries ready for board reports or regulatory submissions — no separate generate step is needed.

Pricing

Deep audits cost £8 per job. Report data is included in every check response at no extra cost. For 847 claims, the total quarterly audit cost is approximately £6,776 — offset by early fraud detection and reduced manual review time.