ShelfEye is an enterprise-grade, architecture-focused GTM pipeline designed to ingest raw retail lead data, algorithmically rank high-yield prospects, and dynamically synthesize role-based sales intelligence.
The engine directly targets accounts exposed to the $1.77T global retail loss driven by shelf blind spots discovered during retail manager field interviews. It programmatically sequences pipelines to secure Phase 1 pilots, accelerating the path toward a $100k ARR milestone while defending a target >3:1 LTV:CAC ratio by contrasting a $45/month SaaS pricing model against quantified $500+/day retail aisle losses.
ShelfEye rejects loose script patterns in favor of a clean, decoupled, production-grade architecture. This ensures strict deterministic mathematical boundaries for lead scoring alongside structural flexibility for semantic analysis.
- Ingestion Layer (Streamlit Blueprint): Designed to handle structured CSV uploads containing core account variables (daily revenue, labor friction status, and GTM pilot readiness).
- Deterministic Scoring Engine (Python/Pandas): Eliminates subjective positioning by executing a rigid, weighted mathematical matrix to calculate a definitive Priority Score (0-100).
- Data Caching & Cost Defense Layer (SQLite Blueprint): Implements a relational relational storage cache to intercept duplicate accounts, mitigating redundant LLM token expenditures to safeguard unit economics.
- Structured GenAI Synthesis Layer (OpenAI API / GPT-4o-mini): Enforces strict JSON schemas to eliminate model drift, generating dual-persona sales briefs that map store-manager operational workflows directly onto corporate buyer margin goals.
ShelfEye/
│
├── src/ # Core Application Source
│ ├── app.py # Streamlit UI Layer Blueprint
│ ├── engine.py # Deterministic Scoring Matrix
│ └── llm_client.py # OpenAI Structured JSON Interface
│
├── .gitignore # Python Environment Hygiene
├── requirements.txt # Dependency Manifest
└── README.md # System Documentation# ShelfEye