Archangel – Convertis

Convertis is a structured AI-enabled decision-support system for commercial real estate screening. The project is better framed as an explainable screening and scorecard platform than as a simple model interface.

The workflow starts with a property address and property type, enriches the request with location context, computes a deterministic commercial feasibility score, supports AI-assisted research, and produces reviewable scorecards and benchmarking outputs.

The public architecture emphasizes bounded inputs, transparent scoring criteria, analyst approval points, reusable evidence, and structured outputs that can support repeatable property review.

• Translated the source technical specification into a public case-study structure with clear architecture, data-flow, engineering notes, and publishable visual assets.
• Separated public product and AI-engineering concepts from internal implementation details so the case study can be shown as portfolio material without exposing private configuration.
• Framed the system around analyst workflows, explainable scoring, source approval, structured scorecards, and comparable-property review.

1. An analyst enters a property address and property type to create a bounded screening request.
2. The system enriches the property with location and nearby-place context suitable for commercial feasibility review.
3. Signal extraction converts location context into comparable criteria such as amenity density, retail concentration, transit proximity, and data confidence.
4. A deterministic scoring layer computes an overall score, grade, criterion breakdown, signal values, and human-readable explanation.
5. AI-assisted research supports evidence collection, source review, and preliminary commercial or technical scorecard drafting.
6. Approved outputs become reviewable scorecards, insights, benchmark comparisons, and property-level artifacts.

{
  "project": "Archangel – Convertis",
  "workflow": "property_screening",
  "input": {
    "property_type": "mixed_use_retail",
    "market": "South Florida",
    "address_status": "geocoded"
  },
  "signals": {
    "amenity_count": 42,
    "retail_concentration": "high",
    "transit_proximity": "medium",
    "data_confidence": 0.82
  },
  "scorecard": {
    "overall_score": 78,
    "grade": "B+",
    "criteria": [
      {
        "name": "commercial_density",
        "score": 84,
        "explanation": "Nearby-place concentration supports customer access."
      },
      {
        "name": "accessibility",
        "score": 71,
        "explanation": "Transit proximity is acceptable but not dominant."
      }
    ]
  },
  "review": {
    "analyst_review_required": true,
    "approved_sources": 5,
    "benchmark_ready": true
  }
}

• A deterministic scoring core is a strong engineering choice because analysts can inspect the signal values, criterion breakdown, grade, explanation, and scoring version behind the final result.
• AI is positioned as a research and synthesis layer, not as the sole decision-maker. The system remains more credible when source approval and scorecard review stay visible.
• Benchmarking gives the workflow a practical business endpoint: the result is not just a generated summary, but a comparable property record that supports repeated review.
• The public case study intentionally excludes private configuration, environment settings, operational controls, and internal security notes.

• Demonstrates the difference between productizing AI and adding an isolated model endpoint.
• Shows how deterministic business logic and AI-assisted research can work together in a governed workflow.
• Highlights practical CRE screening concerns: location context, explainable feasibility scoring, source approval, scorecards, and comparable-property benchmarking.
• Provides a strong flagship case study for AI engineering, workflow design, and structured decision-support systems.