EdgeVision

Two Products · One Platform

One edge appliance.
Two operational lenses.

EdgeVision is the on-prem Vision + GenAI runtime. RetailTrack and RiderTrack are the two products built on top of it — same edge platform, same privacy guarantees, same control plane.

A retail store floor with AI detection overlays on customers and staff

RETAILTRACK

Intelligence for Store Floors & QSR.

Footfall, dwell, VIP recognition, staff coverage, drive-thru wait time, food-safety compliance — every camera becomes a real-time operations sensor for apparel retail and quick-service restaurants alike.

Customer & staff analytics on the same feed
QSR speed-of-service, kitchen choke-point & hygiene
VIP recognition & loss-prevention with zero biometrics off-device

Explore RetailTrack →

A dark-store loading bay at night with delivery riders and AI overlays

RIDERTRACK

Intelligence for the Last Mile.

Idle-cluster detection, presence verification, demand-and-weather correlation. The hub finally sees what the dispatcher can't.

Cluster detection before the surge model fires
Privacy-preserving presence vs roster verification
Helmet, bag, parking & COD-intake compliance

Explore RiderTrack →

PRODUCT 01 · RETAILTRACK

Intelligence for
Store Floors & QSR.

High-performance Intel edge compute on the site LAN. Hybrid Vision + GenAI on-prem. Every camera becomes a customer-and-staff intelligence sensor — across apparel retail floors, supermarkets, and quick-service restaurants — without storing a single face.

RETAILTRACK · CHAPTER 01

The Visibility Gap

Modern apparel retail captures thousands of hours of CCTV every week. Almost none of it becomes a decision.

The Challenge

The In-Store
Visibility Gap.

Stores capture hours of CCTV every shift. But almost none of it becomes a decision.

A dim apparel retail store interior with a dome CCTV camera visible in the upper right corner

01

Staff effort & productivity is opaque

Floor managers can't see who is attending, grouping, or on the phone. Wall locations sit unmanned. Conversion is attributed by gut feel.

02

Customer journey is invisible after entry

VM decisions are made without knowing which entry, gondola, wall, or centre table actually pulls dwell. Footfall ≠ engagement.

03

High-value customers walk in unannounced

Store managers learn about a tier-gold visitor only at the bill counter. Loyalty runs blind to the first 60 seconds.

04

Floor hygiene fails surface late

Footwear scattered, displays disturbed, > 4 staff in back store — visible-but-unmonitored issues that drive complaints and shrink.

RETAILTRACK · CHAPTER 02

The Platform

High-performance Intel edge AI on the store LAN. Hybrid Vision + GenAI on-prem. Insights in seconds.

A compact Intel-powered edge compute device on a counter next to a slim monitor

The Solution

EdgeVision Platform.

High-performance Intel edge compute on the store LAN ingests existing CCTV, runs hybrid Vision + GenAI on-prem, and produces real-time alerts, dashboards, and APIs.

01

Connect Cameras

Existing IP / CCTV via RTSP. No new wiring.

→

02

Process on Edge

Intel edge compute inside store. Air-gapped & isolated.

→

03

Analyse & Detect

Vision + GenAI models reason over the scene.

→

04

Generate Insights

Alerts, KPIs, clips, dashboards, APIs, WhatsApp.

Real-timeDetection

Multi-camPer edge node

0KBCloud bandwidth

<1wkDeployment

Live Operations Console

One Pane of Glass
for the Store Floor.

FOOTFALL1,284

ACTIVE STAFF12 / 14

ALERTS3

AVG DWELL4:21

HOURLY FOOTFALL

10AM2PM6PM10PM

ALERT STREAM

VIP entered · gold tier
Wall #3 unattended 2m
Queue > 4 at till 2
Back store > 4 staff

CAM 01CAM 02CAM 03 CAM 04CAM 05CAM 06

RETAILTRACK · CHAPTER 03

Two Lenses,
One Network

The same feeds answer two questions at once: what is the customer experiencing? And how is the team performing?

A dark apparel store with magenta detection overlays tracking customers and staff

Two Lenses

Customer & Staff Analytics.

CUSTOMER ANALYTICS

What customers do, where they linger, who they are.

Heat-mapping by entry, fixture and category
Walk-in flow between apparel and footwear
Self-service vs assisted journey detection
VIP / loyalty member recognition at entry
POS ↔ camera time-stitch for billing intelligence

STAFF ANALYTICS

Who is attending, who is missing, who is grouping.

Peak vs non-peak presence by zone
Grouping > 1 min, mobile use, back-store time
Style/size run timing — wall to back-store
Per-staff customer attention & conversion proxy
Brand-level heat-map for back-store activity

Same hardware. Same cameras. Two outcomes. Intel edge compute on the store LAN — no extra installation.

In-Store Detection

Tracking Customers
& Staff Simultaneously.

Every person on the floor is detected, classified, and tracked — without ever storing a face.

A live overhead view of a retail store with customers and staff outlined by AI detection boxes

Capability 01

Customer Analytics,
Elaborated.

1

Heat Mapping

By entry, fixture, category. Walk-in flow between categories becomes a measurable cross-category insight.

2

Customer Tracking

VIP & loyalty recognition at entrance. Auto-enroll dwell > 15 min. No biometric data leaves the store.

3

Self-Service vs Assisted

Identifies customers who buy without staff help — critical for VM, range planning and staffing.

4

POS ↔ Camera Stitch

Correlate bills with in-store trail. Every transaction becomes a journey: dwell, attended-by, time-to-bill.

Heat Mapping

Customer Dwell, Visualised.

CONCEPT FLOOR · DWELL HEAT

TOP DWELL ZONES

Brand Wall92%

Centre Table78%

Gondola54%

Entry41%

Cash Desk32%

Generated entirely on the in-store edge box. No cloud round-trip.

VIP Recognition Flow

From CCTV Frame
to Manager's Phone.

A loyalty member walks in. Within 3 seconds the manager has identity, history, and a recommended action — without a single cloud round-trip.

A well-dressed customer entering an upscale store, greeted by a sales associate with a tablet

010.4s

Recognition

Privacy-preserving embedding match against the store-local gallery. No raw faces leave the box.

→

022.1s

Notification

Manager pinged via WhatsApp / app — name, tier, last visit, last basket, preferences.

→

033.0s

Activation

Next-best-action: assign senior associate, push targeted POS offer, trigger follow-up sequence.

Capability 02

Staff Analytics,
Elaborated.

1

Grouping & Inattention

Flags clusters of staff grouped > 1 minute without an active customer. Manager receives a real-time alert with clip.

2

Wall-Location Coverage

Detects when a staffed wall is unattended beyond threshold during peak hours. Trended hourly, store by store.

3

Mobile-Use Tracking

Phone use detected with thresholds (1 min floor, 2 min back store). Coaching tool, not a surveillance tool.

Capability 02 · Continued

The Hidden Loops
That Decide a Sale.

4

Style / Size Run-Time

Time between staff leaving the brand location to look for a size and returning with stock. Per-brand, per-staff, per-shift.

5

Per-Staff Sale Conversion

Stitching the journey to attending staff and POS bill yields an objective attended-conversion score.

A retail store with customer trajectories traced in magenta dotted lines

PER-STAFF ACTIVITY · S04 · SHIFT 10–8

Attended

Wall Station

Back Store

Mobile Use

Grouping

10AM122PM468PM

Brand-Level Staff Heat Map

Where Your Staff
Actually Spend Time.

FLOOR MOVEMENT MAP

Per-staff movement trail across the floor
Brand-zone occupancy heat by 30-min slot
Hand-off detection between brand zones
Comparison: peak vs non-peak coverage

BACK-STORE ACTIVITY

Round-trips per brand per day (size hunts)
Time spent by each staff in back store
Alert if > 4 staff in back store at once
End-of-day back-store occupancy report

RETAILTRACK · QSR EDITION

From Counter to
Drive-Thru.

Quick-service restaurants run on three numbers: speed, accuracy, and food safety. Every camera in the dining room, kitchen, and drive-thru already sees them. RetailTrack turns them into a live signal.

A busy QSR counter and kitchen at evening with AI overlays on staff and customers

QSR Capabilities

Speed, Accuracy,
Safety — Live.

Three operational levers every QSR brand tracks weekly via spreadsheets and mystery-shopper audits. RetailTrack tracks them per-shift, per-station, per-car — automatically.

1

Speed of Service

Drive-thru wait time per lane, counter queue length, mobile-pickup dwell, time-to-first-bag — every daypart benchmarked against itself, not a corporate average.

2

Kitchen Choke-Points

Idle stations, expo bottleneck, fry-station coverage, grill hand-off delays — flagged the moment they exceed threshold, not after the rush has already cost a guest.

3

Food Safety & Hygiene

Gloves, hairnets, handwash frequency, holding-time at the heat lamp, spillage detection — scene-aware checks with low false-positive, ready for the next FSSAI / health audit.

Drive-Thru Operations

Every Lane.
Every Car. Every Second.

The drive-thru is the single biggest revenue line at most QSR sites — and the hardest to measure without a manager standing in the lot. RetailTrack instruments it end-to-end on the cameras you already have.

Live per-lane queue length & wait time
Order-board → handoff-window timing per car
Pull-forward, blocked-lane & "driver not served" alerts
Lane-utilisation balance during peak hours
Daypart vs forecast adherence

A dusk drive-thru lane with cars queued and magenta AI overlays tracking wait times

More Out-of-the-Box · QSR

Every Camera,
Every Service Window.

01

Counter & Kiosk

Queue length vs labour deployed by daypart
Self-order kiosk abandonment & completion rate
Greeter / smile-at-window compliance
Mobile-order pickup-bay dwell & handoff

02

Kitchen & Holding

Station coverage per shift (grill · fry · expo)
Hot-hold & cold-hold zone compliance
Order assembly time & bagging accuracy
Wipe-down cadence between rushes

03

Cleanliness & Safety

Dining-area spillage & tray-bus cadence
Restroom check cadence vs schedule
Slip-and-fall hazard auto-flagging
End-of-day closing-audit photo trail

04

Compliance & Audit

PPE: gloves · hairnet · apron per station
Handwash frequency vs corporate SOP
Camera health & obstruction monitoring
Daily / weekly executive summary email

RETAILTRACK · CHAPTER 04

More Out of the Box

A long tail of operational use-cases — each individually worth the install — running on the same edge appliance.

More Out-of-the-Box

The Long Tail
of Use-Cases.

01

Floor & Display Hygiene

Footwear out of place > 2 min triggers alert
Restore-display task auto-assigned
Daily hygiene compliance score per zone
End-of-day VM delta photos

02

Queue & Cash-Desk

Live queue length & wait time per till
Auto-alert when queue > 4 customers
Bagging / hand-off duration tracking
Trial-room turnover & queue at fitting bay

03

Loss Prevention

Tag-removal & concealment cues
After-hours motion in restricted zones
Unauthorised back-store entries
Door-open / cash-drawer correlation

04

Operations & Compliance

Opening & closing procedure audit
Mannequin / dummy-presence at promo zones
Camera health & obstruction monitoring
Daily / weekly executive summary email

PRODUCT 02 · RIDERTRACK

Intelligence for
the Last Mile.

An AI-powered surveillance and video-summarisation product for quick-commerce dark stores — running on high-performance Intel edge compute that sees what the dispatcher cannot.

A delivery hub at night with riders and magenta AI detection overlays

RiderTrack · The Challenge

The Dark-Store
Visibility Gap.

Quick-commerce dark stores face inefficiencies as delivery agents exploit the system during difficult conditions — rain, traffic, low-demand windows. The dispatcher sees the symptom; nobody sees the cause.

A cluster of idle delivery riders standing together at a hub at night

01

Idle behaviour during difficult conditions

During rain or heavy traffic, some agents deliberately remain idle in waiting areas — falsely signalling a shortage of available riders to the dispatch system.

02

Unverified agent presence

The roster says the rider is on shift. Without privacy-preserving recognition, the hub has no objective record of who is actually on the apron and active.

03

Artificial surge & cost spikes

False scarcity triggers artificial surge pricing and unnecessary cost spikes. Customer charges go up, customer satisfaction goes down — both manufactured by the gap.

04

Operational opacity

Floor managers and HQ have no transparent insight into idle patterns, exploitation, or accountability — until an audit, weeks later. Workforce decisions stay gut-driven.

RiderTrack · The Solution

RiderTrack Platform.

An AI-powered surveillance and video-summarisation product. RiderTrack onboards agents via privacy-preserving recognition, tracks presence and movement, monitors real-time activity, and flags idle behaviour — providing transparent insights to operations teams and enabling timely intervention before the surge model fires.

01