Engineering Blog · Post #10

Unified Document Management: Drag-Drop Upload, LOB Checklists, Multi-Storage (S3/SharePoint/Network), and SHA-256 Dedup

How InsightUW handles a 47-document Construction Builders Risk submission — with LOB-specific checklists, automatic deduplication, and a completeness score that tells the underwriter exactly what is missing before they open a single file.

The Problem

A Construction Builders Risk submission is one of the most document-heavy submissions in commercial insurance. A single submission can include architectural plans, soil reports, contractor prequalification packets, OCIP/CCIP certificates, project timelines, financial statements, loss runs, and more — easily 30 to 50 documents.

In most underwriting workstations, document management is an afterthought:

Documents are attached to emails and manually renamed
There is no checklist telling the underwriter which documents are required vs. received
Duplicate uploads waste storage and confuse reviewers ("Is v2 the latest or the one without track changes?")
Storage is fragmented — some documents live in S3, others in SharePoint, and the really old ones are on a network share
The document management system (DMS) has no idea what the underwriting platform has, and vice versa

The underwriter spends 25–40 minutes per submission just organizing, renaming, and checking document completeness — before they even start underwriting.

The InsightUW Approach

InsightUW provides a unified document management layer with four capabilities that work together: drag-drop upload with automatic classification, LOB-specific required document checklists, multi-storage adapter architecture, and SHA-256 content deduplication.

graph TB subgraph Upload["Upload Interface"] A["Drag & Drop Zone"] B["Bulk File Selector"] C["Email Attachment Auto-Ingest"] D["API Upload (Bot / RPA)"] end subgraph Processing["Document Processing Pipeline"] E["SHA-256 Hash Computation"] F["Duplicate Detection"] G["AI Classification (Document Type)"] H["Metadata Extraction (Page Count, Size)"] end subgraph Storage["Storage Adapters"] I["AWS S3 (Primary)"] J["Share Point (Collaboration)"] K["Network Share (Legacy)"] L["Azure Blob (DR)"] end subgraph Checklist["LOB Checklist Engine"] M["Required Docs per LOB"] N["Completeness Score"] O["Missing Doc Alerts"] end A --> E B --> E C --> E D --> E E --> F F -->|"New"| G F -->|"Duplicate"| P["Reject / Link Existing"] G --> H H --> I H --> J H --> K H --> L G --> M M --> N N --> O

The Document Model

Every document in InsightUW is tracked with a comprehensive metadata record:

LOB-Specific Document Checklists

Each LOB in InsightUW has a configurable document checklist that defines which documents are required, recommended, and optional. The checklist drives the completeness score.

SHA-256 Deduplication Mechanism

Document deduplication is critical when brokers send the same file multiple times across emails, portal uploads, and follow-up correspondence. InsightUW computes a SHA-256 hash of every uploaded file and checks it against all documents in the submission.

sequenceDiagram participant User as Underwriter / Broker participant UI as InsightUW Upload participant Hash as SHA-256 Engine participant DB as Document Store participant Storage as S3 Adapter User->>UI: Drops "Geotechnical_Report_v2.pdf" (4.5MB) UI->>Hash: Compute SHA-256 hash Hash-->>UI: a7ffc6f8bf1ed76...f8434a UI->>DB: Select * From documents Where sha256 hash = 'a7ffc6f8...' AND submission id = 'SUB-2026-04-1847' alt Hash Match Found DB-->>UI: Existing doc: "Geotechnical_Report_Riverside_Tower.pdf" (uploaded 2 days ago) UI-->>User: ⚠️ Duplicate detected. "This file is identical to Geotechnical_Report_Riverside_Tower.pdf uploaded on Apr 18. Link to existing or upload anyway?" User->>UI: "Link to existing" UI->>DB: Create document link record (no new storage used) else No Match DB-->>UI: No duplicate found UI->>Storage: Upload to S3 Storage-->>UI: storage path confirmed UI->>DB: Insert document record with hash UI-->>User: ✅ Upload complete. Classified as "Soil / Geotechnical Report" end

The dedup API is also available for programmatic checks:

Multi-Storage Adapter Architecture

InsightUW does not mandate a single storage backend. A storage adapter interface allows the platform to read and write documents across multiple backends transparently.

graph LR subgraph Interface["Storage Adapter Interface"] A["upload(file, path)"] B["download(path)"] C["delete(path)"] D["generate presigned url(path, ttl)"] E["list(prefix)"] end subgraph Adapters["Concrete Adapters"] F["S3Adapter Primary cloud storage"] G["Share Point Adapter Office 365 integration"] H["Network Share Adapter Legacy UNC paths"] I["Azure Blob Adapter DR / multi-cloud"] end subgraph Config["Routing Rules"] J["LOB → Adapter mapping"] K["File size routing (>50MB → S3 only)"] L["Compliance routing (PII → encrypted tier)"] end Interface --> F Interface --> G Interface --> H Interface --> I Config --> Interface

The Scenario

Willis Towers Watson submits a Builders Risk application for the Riverside Tower project in Miami — a 42-story luxury residential tower with a $280M total project value. The submission arrives via email with 12 initial documents, followed by 6 more uploaded through the broker portal, and 3 duplicates re-sent in a follow-up email.

What Happens in InsightUW (Timeline)

Time	Event	System Action
10:00 AM Mon	Email received with 12 attachments	Mailbox monitor triggers; 12 files queued for processing
10:00:04 AM	SHA-256 hashes computed for all 12 files	All 12 are unique; no duplicates in the system
10:00:06 AM	AI classification runs on each document	ACORD App (0.97), Architectural Plans (0.95), Project Budget (0.91), Loss History (0.93), GC Prequal (0.89), Soil Report (0.94), Project Timeline (0.88), Permits (0.92), Safety Program (0.86), Site Photos (0.82), Financial Stmts (0.90), Loss Runs (0.91)
10:00:08 AM	Documents stored in S3	12 files → `submissions/2026/04/SUB-2026-04-1847/` organized by category
10:00:08 AM	Checklist evaluated	10 of 18 required/recommended docs received. Completeness: 67%
10:00:09 AM	Missing doc alert generated	Missing: Supplemental App, Structural Engineering, Subcontractor List, Environmental Assessment, OCIP/CCIP Cert, Current Policy Dec
2:15 PM Mon	Broker uploads 6 more docs via portal	Drag-drop upload; 6 files hashed and classified
2:15:03 PM	Checklist re-evaluated	16 of 18 docs received. Completeness: 89%
9:00 AM Tue	Broker re-sends 3 docs from Monday's email	SHA-256 match on all 3. Duplicates rejected with link to existing docs
9:00:01 AM	Storage savings logged	14.2 MB of duplicate storage avoided
9:30 AM Tue	Final 2 docs uploaded (Environmental Assessment, OCIP cert)	Completeness: 100% — all required and recommended documents received

DMS Integration

For carriers that already have a Document Management System (FileNet, OnBase, Laserfiche), InsightUW provides a bidirectional sync layer:

Metrics: Before and After InsightUW Document Management

Metric	Before InsightUW	After InsightUW	Improvement
Time to organize submission documents	25–40 min per submission	0 min (auto-classified)	100% eliminated
Document completeness visibility	Manual checklist in Excel	Real-time score per submission	Instant
Duplicate storage waste	15–25% of total storage	< 1% (SHA-256 dedup)	95% reduction
Missing document discovery	Days (underwriter manually reviews)	Seconds (auto-checklist)	99% faster
Broker follow-up for missing docs	Manual email after review	Auto-generated missing doc list	Automated
Storage cost (annual)	$120K (duplicates + fragmentation)	$78K (deduplicated + tiered)	35% savings
Document retrieval time	30–90 sec (search across systems)	2–5 sec (unified index)	95% faster
DMS sync accuracy	70% (manual filing)	99.8% (automated classification)	30% improvement

Key Takeaways

Document management is underwriting infrastructure, not an afterthought. InsightUW treats every document as a first-class entity with classification, dedup, and checklist awareness.
LOB-specific checklists eliminate the "what's missing?" question. The completeness score tells the underwriter and the broker exactly what is needed before the review even begins.
SHA-256 dedup saves storage and eliminates confusion. No more "which version is the latest?" — identical files are detected and linked, not re-stored.
Multi-storage adapters meet you where your documents live. S3, SharePoint, network shares, Azure Blob — InsightUW reads and writes across all of them through a single interface.
DMS integration is bidirectional. Documents flow from InsightUW to your DMS and back, maintaining a single source of truth regardless of where the document physically resides.

Ready to unify your document chaos? InsightUW's document management layer handles the 50-document Builders Risk submission as easily as a single ACORD application — with dedup, checklists, and multi-storage built in.

Schedule a Document Management Demo →