Food Safety · Edge AI · Quick Commerce

HubSentinel

AI quality gate for quick commerce dark stores

Catch spoiled produce, expired stock, and vendor violations at intake — automatically, offline, before they reach a customer.

The Problem

📦

Spoiled stock ships

Manual intake checks are inconsistent. A tired picker misses a soft tomato. A bad batch gets shelved. The customer gets it, returns it, and leaves a review.

📅

Expiry dates go unchecked

High SKU volumes make it impossible to verify every expiry date by hand. Near-expired stock gets dispatched. FSSAI liability follows.

🏭

Vendor accountability is missing

When a quality complaint comes in, there's no record of which vendor batch it came from, who accepted it, or when. Root cause is impossible.

How HubSentinel Works

A camera, an edge device, and a dashboard — deployed at the intake point of your dark store

01

Camera captures produce at intake

An overhead camera films items as they arrive. No conveyor belt modifications, no special lighting rigs required.

02

On-device AI grades freshness in real time

A MobileNetV2 model running on a Raspberry Pi 5 classifies freshness in under 30ms per item — no internet connection needed, no data sent to the cloud.

03

OCR reads expiry dates automatically

Packaged goods are scanned for printed expiry dates. Items flagged as near-expired or expired are logged and held.

04

Rejections logged by vendor batch

Every rejection is recorded against the vendor and batch — exportable in FSSAI-compatible format for audit readiness and supplier accountability.

Key Features

Fully Offline

Runs entirely on a Raspberry Pi 5 at the edge. No cloud dependency, no latency, no data leaving your premises. Works in dark stores with poor connectivity.

High Accuracy Inference

MobileNetV2 model trained on fresh/spoiled produce categories with over 99% validation accuracy. 22–31ms inference per item.

FSSAI-Format Export

Rejection records export directly in FSSAI-compatible CSV format. Audit-ready at any time without manual data compilation.

Live Dashboard

Browser-based dashboard with live camera feed, item-level pass/fail status, vendor batch logs, and rejection trend charts.

Vendor Accountability

Every rejection is tied to a vendor and batch ID. Build a quality scorecard across vendors over time and renegotiate or delist on data.

Minimal Setup

No infrastructure changes needed. Plug in a camera, connect to local Wi-Fi for the dashboard, and it's running. AP mode available for fully isolated deployments.

Who It's For

Request a Demo

We'll walk you through a live demo and assess fit for your operation — no commitment required.

Looking for something different? Tell us about your project →