Permits, Inspections
& Evidence
PIE is a property-level regulatory-data extraction agent in the InsureMEP ecosystem. It pulls permits, inspections, violations, ownership, and evidence from 60+ public agencies, normalizes everything around the property + asset graph, and feeds CriticalAsset.
What is PIE?
Asset-First Architecture
Properties are shells. The real intelligence lives in the asset graph — systems, equipment, relationships, dependencies, and lifecycle events that determine operational readiness and risk.
Gemini AI Intelligence
Every permit is enriched by Google Gemini into structured intelligence — systems impacted, asset classes identified, lifecycle events inferred, insurance signals scored, and operational relevance assessed.
Insurance-Native Signals
Purpose-built for InsureMEP underwriting workflows. Every data point carries compliance relevance, insurance relevance, and operational relevance scores with human-in-the-loop verification.
Core Capabilities
📋 Permit Intelligence
Ingest raw permits from any jurisdiction. Normalize addresses, types, and contractors. Run AI enrichment to extract systems, assets, and lifecycle signals from work descriptions.
- Multi-jurisdiction permit ingestion
- Address normalization and parcel matching
- AI-powered work description analysis
- Contractor activity correlation
🔍 Inspection Tracking
Track inspection events, results, and findings across the entire property portfolio. Automatically generate findings and obligations from failed inspections.
- Pass/fail/partial tracking per inspection type
- Inspector assignment and contact management
- Automatic finding generation from failures
- Follow-up obligation creation
📊 Evidence Management
Collect, categorize, and score evidence records. Track commissioning reports, test results, certificates, photos, and warranty documents with strength and relevance scoring.
- Evidence strength classification (strong/moderate/weak)
- Compliance, insurance, and operational relevance
- AI confidence scoring
- Human verification workflow
⚡ Insurance Signal Generation
Derive underwriting-oriented risk signals from permit and inspection intelligence. Every signal carries direction (positive/negative), severity, domain, and confidence.
- Equipment upgrade / capacity change detection
- Failed inspection and compliance gap signals
- Deferred maintenance indicators
- Emergency repair and major renovation tracking
Operational Modules
| Module | Description | Key Data Points |
|---|---|---|
| Jurisdictions | Government entities that issue permits | Name, state, type, data source URL, active status |
| Properties | Commercial buildings and facilities | Address, parcel ID, type, year built, sq ft, floors, owner, class |
| Locations | Specific areas within properties | Name, floor, zone, location type |
| Systems | Building systems (HVAC, electrical, fire, etc.) | Name, type, domain, criticality, install year |
| Assets | Individual equipment and components | Name, type, manufacturer, model, serial #, condition, criticality |
| Contractors | Licensed service providers | Company, license #, license type, contact info, active |
| Permits | Building permits from any jurisdiction | Permit #, address, type, work description, status, valuation |
| Inspections | Inspection events and results | Type, date, inspector, result, notes |
| Findings | Deficiencies, violations, gaps | Title, type, severity, domain, status, due date |
| Evidence | Reports, certificates, test results | Title, type, strength, compliance/insurance/ops relevance |
| Insurance Signals | Underwriting risk indicators | Signal type, direction, severity, domain, confidence |
| Obligations | Required follow-up actions | Title, type, domain, priority, status, due date |
| Lifecycle Events | Asset/system lifecycle milestones | Event type, date, summary, confidence, review status |
| Relationships | Entity-to-entity connections | Source/target entities, relationship type, confidence |
| Market Targeting | Outreach priority scoring | Permit density, systems complexity, criticality, insurance value |
✦ AI Auto-Populate
From Simple Clue to Structured Intelligence
Every create form in PIE includes an AI Auto-Populate bar. Type a simple clue — a building name, an equipment description, a company name — and Gemini fills all form fields with real-world data.
Supported across all 13 entity types: properties, assets, systems, permits, jurisdictions, contractors, inspections, findings, evidence, insurance signals, obligations, lifecycle events, and locations.
Data Architecture
Technology Stack
Frontend
- Next.js 14 (App Router)
- React Server Components
- Custom CSS Design System
- Inter Typography
Backend
- Next.js API Routes
- PostgreSQL 16
- JWT Authentication
- bcrypt Password Hashing
Intelligence
- Google Gemini 2.0 Flash
- 13-Entity AI Auto-Populate
- Structured Enrichment Pipeline
- Human-in-the-Loop Review