Insurance claim assessment and compliance report from supporting documents

Introduction Insurance claim adjusters spend hours reviewing supporting documents, cross-referencing policy details, and writing assessment reports that follow strict compliance templates.

Each claim requires careful extraction of key facts, date verification, and standardised documentation. When adjusters process dozens of claims weekly, this manual work becomes a significant bottleneck. The work is repetitive, error-prone, and keeps skilled staff away from decisions that actually need human judgment. The good news: most of this workflow can be automated. You can build a system that ingests claim documents, extracts relevant information, generates compliant assessment reports, and routes them for final review without anyone manually copying and pasting between tools or retyping information. This Alchemy recipe shows you how to connect PDF document analysis, content summarisation, and report generation into a single automated workflow using n8n as your orchestration platform. When a claim document arrives, the system will read it, extract key facts, summarise supporting evidence, and produce a formatted compliance report ready for sign-off.

The Automated Workflow Why n8n for this workflow We're using n8n because it has solid webhook support, can handle document uploads directly, and connects cleanly to the tools we need without excessive API wrangling.

Zapier could work too, but n8n gives you more control over request formatting and conditional logic without hitting plan limits as quickly. Architecture overview The workflow operates in four stages: document ingestion via webhook, content extraction using Chat With PDF, summarisation via Smmry, and report generation through Mintlify. Human review happens before final output.

Step 1: Set up the webhook trigger in n8n

Create a new workflow in n8n and add a Webhook node as your entry point. This webhook will receive incoming claim documents from your system or a form submission.

CopyPOST /webhook/claim-intake
Content-Type: multipart/form-data { "claim_id": "CLM-2024-001847", "claimant_name": "Sarah Mitchell", "document_file": <binary PDF data>, "policy_number": "POL-789456", "claim_date": "2024-03-15"
}

Configure the webhook to accept POST requests. Set the output data mapping so that n8n captures the claim_id, claimant_name, and document_file in accessible variables. You'll reference these throughout the workflow.

Step 2: Upload and analyse with Chat With PDF

After the webhook captures the document, pass it to Chat With PDF for content extraction. Chat With PDF supports batch processing of multiple documents and maintains context across questions, which is valuable when you need to cross-reference information across claim documents. Create an HTTP Request node in n8n configured like this:

CopyPOST https://api.chatwithpdf.com/v1/extract
Headers: Authorization: Bearer YOUR_CHATWITHPDF_API_KEY Content-Type: application/json Body:
{ "documents": [ { "content": "{{$node.Webhook.json.document_file}}", "filename": "{{$node.Webhook.json.claim_id}}_support.pdf" } ], "queries": [ "What is the date of loss?", "Who is the policyholder?", "What is the claimed amount?", "What type of damage or loss is claimed?", "List all medical provider names mentioned", "What is the incident location?", "Are there any policy exclusions referenced?" ]
}

Chat With PDF will return extracted answers for each query. Store these responses in n8n variables for the next stage. The tool handles OCR internally, so even scanned PDFs will work.

Step 3: Summarise evidence with Smmry

Now that you've extracted key facts, feed the supporting documentation through Smmry to generate a concise summary of evidence. This serves two purposes: it confirms the extracted facts are correct, and it creates a quotable summary for your compliance report. Add another HTTP Request node:

CopyPOST https://api.smmry.com/summarise
Headers: Authorization: Bearer YOUR_SMMRY_API_KEY Content-Type: application/json Body:
{ "document_text": "{{$node['Chat With PDF'].json.extracted_content}}", "summary_length": 5, "preserve_keywords": ["policy exclusion", "deductible", "coverage limit"]
}

Smmry returns a condensed summary and can highlight specific keywords relevant to compliance. Set summary_length to 5 to keep reports under 200 words.

Step 4: Generate compliance report with Mintlify

With extracted facts and a verified summary in hand, now create the formatted assessment report using Mintlify. Mintlify is purpose-built for structured documentation generation, which means your reports will follow consistent templates and comply with regulatory requirements. Add an HTTP Request node for Mintlify:

CopyPOST https://api.mintlify.com/v1/generate
Headers: Authorization: Bearer YOUR_MINTLIFY_API_KEY Content-Type: application/json Body:
{ "template": "insurance_claim_assessment", "data": { "claim_id": "{{$node.Webhook.json.claim_id}}", "claimant_name": "{{$node.Webhook.json.claimant_name}}", "policy_number": "{{$node.Webhook.json.policy_number}}", "claim_date": "{{$node.Webhook.json.claim_date}}", "extracted_facts": "{{$node['Chat With PDF'].json.extracted_answers}}", "evidence_summary": "{{$node.Smmry.json.summary}}", "assessment_type": "preliminary", "generated_at": "{{now()}}" }, "output_format": "pdf"
}

Mintlify will format everything into a professional PDF report. You can define custom templates within Mintlify to match your organisation's compliance requirements and branding.

Step 5: Route to review and approval

Add a final node that stores the generated report and sends a notification to the assigned adjuster. This preserves the human review step, which is critical for claims work.

CopyPOST https://api.slack.com/api/chat.postMessage
{ "channel": "#claims-review", "text": "New claim assessment ready for review", "blocks": [ { "type": "section", "text": { "type": "mrkdwn", "text": "*Claim ID:* {{$node.Webhook.json.claim_id}}\n*Claimant:* {{$node.Webhook.json.claimant_name}}\n*Report:* {{$node.Mintlify.json.report_url}}" } } ]
}

Conditional error handling Add a Switch node after Chat With PDF to catch extraction failures. If the document is unreadable or missing critical information, route to a manual review queue rather than pushing incomplete data downstream.

The Manual Alternative If you prefer more control over individual steps, you can run this workflow manually in stages.

Download the PDF, open it in Chat With PDF directly, copy extracted facts into a text document, run that through Smmry's web interface, and then generate your report in Mintlify. This takes longer but gives you granular oversight of each decision point. You might choose this approach for high-value or unusually complex claims where you want to verify every extraction yourself before automation proceeds.

Pro Tips Rate limits and batching Chat With PDF and Smmry both have per-minute request limits on their free plans.

If you're processing multiple claims simultaneously, batch them in groups of three to five per minute rather than firing all requests at once. Configure n8n's Rate Limit node to throttle webhook ingestion and spread load evenly. Cost optimisation with model selection Mintlify uses GPT-4o by default for report generation. If you're processing hundreds of claims monthly, test whether GPT-4o mini produces adequate report quality for your use case. The cost difference is significant at volume. Document format and preprocessing Chat With PDF works with PDFs, images, and scanned documents, but performance varies. If you're receiving scanned forms with poor image quality, consider running them through a document enhancement service first. This costs extra but reduces extraction errors and downstream corrections. Monitoring extraction accuracy Spot-check extracted facts against source documents weekly. Log mismatches in a spreadsheet and update your Chat With PDF queries if you notice patterns. For example, if the system consistently misses policy exclusions written in small print, revise the query to say "Look for any mention of exclusions, even in fine print or footnotes." Compliance audit trail Store all extracted data, summarisations, and generated reports in a timestamped audit log. This becomes very useful during regulatory reviews. Include the raw API responses, not just the final report, so you can trace how conclusions were reached.

Cost Breakdown