Building the Future of Retail Intelligence
We're a team of engineers, data scientists, and retail experts building the operating system for store-level truth.
Mission
Move retail tracking from subjective field notes to repeatable, schema-first measurements with provenance, continuous improvement, and a path to portfolio-wide SKU intelligence.
Vision
A world where every product on every shelf is tracked in real time — brand owners, distributors, and retailers share a single source of truth about what's happening at retail.
Approach
We don't just apply AI to retail. We build precision systems: dual-image prompting, profile-specific schemas, arbitrated truth layers, and human-in-the-loop learning that gets better with every survey.
Our Milestones
From a detection prototype to a full retail intelligence platform.
Foundation
Core detection pipeline built — YOLO + BYTETrack + best-shot selection.
Classification
MobileNet v4 multi-label classifier achieving F1 > 0.92 across 8 categories.
Extraction Pipeline
Vision LLM extraction with 7 profile-specific JSON schemas and dual-image prompting.
Truth Layer
Field arbitration engine with OCR, consensus, and external lookup provenance.
SKU Master DB
Persistent brand/product/variant graph with consensus pricing and observation store.
Platform Launch
Full platform with dashboard, analytics, multi-model benchmarking, and enterprise APIs.