Deepread Insurance

DeepRead Insurance Document Processing

Extract structured data from insurance claims, policy documents, Explanations of Benefits (EOBs), loss reports, and adjuster notes. Then redact claimant PII before sharing with third parties.

This skill instructs the agent to POST documents to https://api.deepread.tech and poll for results. No system files are modified.

What You Get Back

Submit a claim form and get structured JSON. Extracted fields come back as a list under extraction.fields[] (each field has key, value, needs_review, and location.page):

{
  "schema_version": "dp02",
  "status": "completed",
  "extraction": {
    "fields": [
      {"key": "claim_number", "value": "CLM-2026-078432", "needs_review": false, "location": {"page": 1}},
      {"key": "policy_number", "value": "POL-HO3-445521", "needs_review": false, "location": {"page": 1}},
      {"key": "claimant_name", "value": "Robert Johnson", "needs_review": false, "location": {"page": 1}},
      {"key": "date_of_loss", "value": "2026-02-14", "needs_review": false, "location": {"page": 1}},
      {"key": "date_reported", "value": "2026-02-15", "needs_review": false, "location": {"page": 1}},
      {"key": "loss_type", "value": "Water Damage", "needs_review": false, "location": {"page": 1}},
      {"key": "loss_description", "value": "Burst pipe in basement caused flooding to finished living area, damaged drywall, carpet, and personal property", "needs_review": false, "location": {"page": 2}},
      {"key": "property_address", "value": "789 Elm Dr, Denver, CO 80202", "needs_review": false, "location": {"page": 1}},
      {"key": "coverage_type", "value": "Homeowners HO-3", "needs_review": false, "location": {"page": 1}},
      {"key": "deductible", "value": 1000.00, "needs_review": false, "location": {"page": 1}},
      {"key": "estimated_damages", "value": 24500.00, "needs_review": true, "review_reason": "Multiple estimates on different pages", "location": {"page": 1}},
      {"key": "adjuster", "value": "Sarah Martinez, License #ADJ-44521", "needs_review": false, "location": {"page": 3}},
      {"key": "line_items", "value": [
        {"category": "Structural", "description": "Drywall replacement — basement", "amount": 8500.00},
        {"category": "Flooring", "description": "Carpet removal and replacement", "amount": 6200.00},
        {"category": "Personal Property", "description": "Damaged furniture and electronics", "amount": 5800.00},
        {"category": "Mitigation", "description": "Water extraction and drying", "amount": 4000.00}
      ], "needs_review": false, "location": {"page": 3}},
      {"key": "status", "value": "Under Review", "needs_review": false, "location": {"page": 1}}
    ]
  }
}

Fields with needs_review: true need human review (check review_reason). Everything else is high-confidence.

Setup

Get Your API Key

open "https://www.deepread.tech/dashboard/?utm_source=clawhub"

Save it:

export DEEPREAD_API_KEY="sk_live_your_key_here"

Insurance Claim Schema

Pre-built schema for insurance claims and loss reports:

{
  "type": "object",
  "properties": {
    "claim_number": {"type": "string", "description": "Claim number or reference ID"},
    "policy_number": {"type": "string", "description": "Insurance policy number"},
    "claimant_name": {"type": "string", "description": "Name of the claimant or insured"},
    "date_of_loss": {"type": "string", "description": "Date the loss or incident occurred (YYYY-MM-DD)"},
    "date_reported": {"type": "string", "description": "Date the claim was reported (YYYY-MM-DD)"},
    "loss_type": {"type": "string", "description": "Type of loss (Water Damage, Fire, Theft, Auto Collision, Liability, etc.)"},
    "loss_description": {"type": "string", "description": "Detailed description of the loss or incident"},
    "property_address": {"type": "string", "description": "Address of the property or location of incident"},
    "coverage_type": {"type": "string", "description": "Type of coverage (Homeowners, Auto, Commercial, Liability, etc.)"},
    "deductible": {"type": "number", "description": "Policy deductible amount"},
    "estimated_damages": {"type": "number", "description": "Total estimated damage amount"},
    "adjuster": {"type": "string", "description": "Claims adjuster name and license number"},
    "line_items": {
      "type": "array",
      "items": {
        "type": "object",
        "properties": {
          "category": {"type": "string", "description": "Damage category (Structural, Personal Property, Medical, etc.)"},
          "description": {"type": "string", "description": "Description of damage or expense"},
          "amount": {"type": "number", "description": "Cost or estimate amount"}
        }
      },
      "description": "Itemized list of damages or expenses"
    },
    "status": {"type": "string", "description": "Claim status (Filed, Under Review, Approved, Denied, Settled)"},
    "settlement_amount": {"type": "number", "description": "Final settlement amount if resolved"},
    "denial_reason": {"type": "string", "description": "Reason for denial if applicable"}
  }
}

Extract Data From Insurance Documents

Python

import requests
import json
import time

API_KEY = "sk_live_YOUR_KEY"
BASE = "https://api.deepread.tech"
headers = {"X-API-Key": API_KEY}

schema = json.dumps({
    "type": "object",
    "properties": {
        "claim_number": {"type": "string", "description": "Claim number"},
        "policy_number": {"type": "string", "description": "Policy number"},
        "claimant_name": {"type": "string", "description": "Claimant name"},
        "date_of_loss": {"type": "string", "description": "Date of loss (YYYY-MM-DD)"},
        "loss_type": {"type": "string", "description": "Type of loss"},
        "loss_description": {"type": "string", "description": "Description of the loss"},
        "coverage_type": {"type": "string", "description": "Coverage type"},
        "deductible": {"type": "number", "description": "Deductible amount"},
        "estimated_damages": {"type": "number", "description": "Total estimated damages"},
        "line_items": {
            "type": "array",
            "items": {"type": "object", "properties": {
                "category": {"type": "string"},
                "description": {"type": "string"},
                "amount": {"type": "number"}
            }},
            "description": "Itemized damages"
        },
        "status": {"type": "string", "description": "Claim status"}
    }
})

with open("claim.pdf", "rb") as f:
    job = requests.post(
        f"{BASE}/v1/process",
        headers=headers,
        files={"file": f},
        data={"schema": schema},
    ).json()

job_id = job["id"]
print(f"Processing: {job_id}")

delay = 5
while True:
    time.sleep(delay)
    result = requests.get(f"{BASE}/v1/jobs/{job_id}", headers=headers).json()

    if result["status"] == "completed":
        fields = result.get("extraction", {}).get("fields", [])
        print(json.dumps(fields, indent=2))

        for f in fields:
            if f.get("needs_review"):
                print(f"\n  REVIEW: {f['key']} — {f.get('review_reason')}")
        break
    elif result["status"] == "failed":
        print(f"Failed: {result.get('error')}")
        break

    delay = min(delay * 1.5, 15)

cURL

curl -s -X POST https://api.deepread.tech/v1/process \
  -H "X-API-Key: $DEEPREAD_API_KEY" \
  -F "file=@claim.pdf" \
  -F 'schema={"type":"object","properties":{"claim_number":{"type":"string","description":"Claim number"},"policy_number":{"type":"string","description":"Policy number"},"claimant_name":{"type":"string","description":"Claimant name"},"date_of_loss":{"type":"string","description":"Date of loss"},"loss_type":{"type":"string","description":"Type of loss"},"estimated_damages":{"type":"number","description":"Total damages"},"status":{"type":"string","description":"Claim status"}}}'

Redact Claimant PII Before Sharing

Redact claimant personal information before sending to adjusters, contractors, or third-party reviewers:

# Step 1: Extract the claim data you need
with open("claim.pdf", "rb") as f:
    extract_job = requests.post(
        f"{BASE}/v1/process",
        headers=headers,
        files={"file": f},
        data={"schema": schema},
    ).json()

# Step 2: Redact PII from the original
with open("claim.pdf", "rb") as f:
    redact_job = requests.post(
        f"{BASE}/v1/pii/redact",
        headers=headers,
        files={"file": f},
    ).json()

redact_id = redact_job["id"]
delay = 5
while True:
    time.sleep(delay)
    result = requests.get(f"{BASE}/v1/pii/{redact_id}", headers=headers).json()
    if result["status"] == "completed":
        report = result["report"]
        print(f"Redacted {report['total_redactions']} PII instances")
        for pii_type, info in report["pii_detected"].items():
            print(f"  {pii_type}: {info['count']} found")
        pdf = requests.get(result["redacted_file_url"]).content
        with open("claim_redacted.pdf", "wb") as f:
            f.write(pdf)
        print("Saved: claim_redacted.pdf")
        break
    elif result["status"] == "failed":
        print(f"Failed: {result.get('error')}")
        break
    delay = min(delay * 1.5, 15)

Use Cases

• Claims Processing — Extract claim numbers, dates, damage descriptions, and amounts from incoming claims
• Policy Document Analysis — Pull coverage terms, limits, deductibles, and exclusions from policy documents
• EOB Processing — Extract procedure codes, allowed amounts, patient responsibility from Explanations of Benefits
• Loss Reports — Parse adjuster field reports for damage categories, estimates, and recommendations
• Subrogation — Extract third-party liability information and recovery amounts
• Fraud Detection — Batch-process claims and flag inconsistencies in dates, amounts, or descriptions
• Auto Claims — Extract vehicle info, driver details, accident descriptions, and repair estimates
• Workers Comp — Pull injury descriptions, medical provider info, and lost wage calculations

Tips for Insurance Documents

• Specify claim-specific field names — Using "Claim number or reference ID" works better than just "number"
• Include status values — Adding expected statuses (Filed, Under Review, Approved, Denied) in descriptions helps extraction
• Create blueprints for recurring form types — If you process the same carrier's claim forms repeatedly, train a blueprint at deepread.tech/dashboard/optimizer for 20-30% improvement
• Always redact before external sharing — Use PII redaction before sending to contractors, adjusters, or legal teams

BYOK — Zero Processing Costs

Connect your own OpenAI, Google, or OpenRouter key via the dashboard. All document processing routes through your provider — zero DeepRead LLM costs, page quota skipped.

Set it up: https://www.deepread.tech/dashboard/byok

Related DeepRead Skills

• deepread-ocr — General OCR and structured extraction — clawhub install uday390/deepread-ocr
• deepread-pii — Redact PII from any document — clawhub install uday390/deepread-pii
• deepread-form-fill — Fill PDF forms with AI vision — clawhub install uday390/deepread-form-fill
• deepread-invoice — Invoice and receipt processing — clawhub install uday390/deepread-invoice
• deepread-medical — Medical records processing — clawhub install uday390/deepread-medical
• deepread-legal — Legal document processing — clawhub install uday390/deepread-legal
• deepread-agent-setup — OAuth device flow authentication — clawhub install uday390/deepread-agent-setup
• deepread-byok — Bring Your Own Key setup — clawhub install uday390/deepread-byok

Support

• Dashboard: https://www.deepread.tech/dashboard
• Demo Repo: https://github.com/deepread-tech/deepread-demo
• n8n Node: https://www.npmjs.com/package/n8n-nodes-deepread
• Issues: https://github.com/deepread-tech/deep-read-service/issues
• Email: support@deepread.tech

Get started free: https://www.deepread.tech/dashboard/?utm_source=clawhub