PRODUCT

AI visual inspection for production lines — Edge appliance + Cloud console.

Name: VIAI
Brand: VIAI by EvoNexon

One product, two parts that work together. Edge inspects on the line. Cloud manages, analyses, and keeps it improving.

01VIAI EdgeOn the line 02VIAI CloudAnalysis & management

By design

Specifications, not slogans.

speed

<50ms

Per-part decision

Fits any line speed.

photo_library

30images

To start training

Good parts only. No labels.

rocket_launch

1day

Install to live

Live by end of shift.

cloud_off

100%

Runs offline

Cloud optional, never required.

01VIAI EDGE · ON THE LINE

AI inspection that runs on your line.

A clear verdict per unit. Reject signal to your line. Fully offline.

bolt

Real-time verdict

Sub-50 ms decision, every unit.

hub

Multi-defect fusion

Sees every defect, weighs each by severity, gives one verdict.

cloud_off

Offline by design

Line keeps running without cloud.

settings_input_component

Plugs into your line

Reject signal, PLC, MQTT.

edge.evonexon.local

movie

Edge Inspector demo

Video coming soon

check_circle

PASS

Part continues.

error_outline

REVIEW

Image saved, sent to a human.

cancel

REJECT

Auto-rejected with evidence.

Multi-defect detection

Many defects. One clear call.

Real parts rarely fail in just one place. We weigh every defect together and call it the way a seasoned inspector would.

Detected on this unit

minorsurface scratch0.46

majorlow fill level0.84

criticalcracked weld0.91

hubSEVERITY FUSION

Final decision

cancelDEFECT

FUSED 0.94

Critical defect → reject regardless

check_circle

1 minor → PASS

Within tolerance.

error_outline

2+ minors → REVIEW

Send to a human.

cancel

Any critical → REJECT

Always rejected.

Tools & integration

Tune, dry-run, plug in.

A workbench for your team, a simulator for safe testing, broad camera support.

science

QA workbench

Inspection Studio

Your QA team tunes the inspector against real parts. Find the right balance, on your data.

bar_chart Live performance charts
compare Compare good vs. defective batches
layers See each step's contribution

play_circle

Dry-run mode

Simulator

Test the inspector before you touch the line. Stream from saved images instead of the camera.

speed Match your line speed
tune Set the defect rate you want
shuffle Replay any scenario

videocam

Camera I/O

Bring your own cameras

Use what you already have. USB, RTSP, and IP cameras supported, with multi-camera rigs up to 4.

usb USB3 industrial cameras
lan RTSP & IP cameras
wb_sunny Auto-exposure & white-balance controls

The edge appliance

One small box.
Sits next to your line.

Compact and low-power. No server room. No GPU rack. No cloud round-trips.

Footprint

Fits in a control cabinet

Install

DIN rail, wall, or enclosure

Operating Range

Industrial-grade

Camera I/O

USB 3 / GigE

Line Output

PLC / MQTT

Connectivity

Wired or wireless

Inside the box

Camera in. Verdict out. Under 50 ms.

Cloud sync is optional - the line keeps running either way.

Data Pipeline

Capture → Infer → Decide → Sync

42 ms END-TO-END

videocamCamera

USB3 / GigE industrial camera mounted on the line.

FPS14.3EXP4.2ms

arrow_forward

memoryEdge Inference

Inspection model runs on the on-site edge appliance.

LAT42 msRAM2.1 GB

arrow_forward

ruleVerdict + Action

PLC trigger, reject arm, alert log - all driven from edge.

OUT24V GPIOAPIREST/MQTT

cloud_syncOptional: Cloud Dashboard syncs results across all your sites for fleet analytics. The line keeps running even when the network is down.

Pick the rig that fits your line.

Four configurations. One for every line.

Configuration

Solo

Every line, every site, one screen.

Live yield. Drift alerts. One-click audit reports.

dashboard

One pane of glass

Every line, every site, live.

notifications_active

Alerts that find you

Slack, email, pager.

history

Audit-ready

Every verdict, replayable.

app.evonexon.com

movie

Cloud Console demo

Video coming soon

For managers

A console that turns inspection into decisions.

The edge catches defects in 42ms. The cloud tells you which line, which shift, which die - and what's about to drift before it costs you scrap. Built for QA leads, plant managers, and the people they report to.

lockapp.evonexon.com / fleetSample data

CONSOLE

Fleet overview

Units inspected · 24h

88,544

trending_up+4.2% vs yesterday

First-pass yield

99.3%

trending_up+0.4 pts WoW

Defects caught

849

trending_down−12% vs avg

$ scrap saved · MTD

$284K

trending_upvs $186K last month

Return14.2×this quarter

All lines

8 active · 1 offline · live

LineYieldUnits · 24hDefectsΔ

Bangalore · A199.2%12,847102+0.3

Bangalore · A298.8%11,203134−0.1

Chihuahua · B196.4%9,482341−1.4

Chihuahua · B299.0%8,93089+0.2

Prague · C199.6%14,00856+0.4

Prague · C2----

Osaka · D199.8%16,20432+0.1

Osaka · D299.4%15,87095+0.2

Active alerts3

warning

Yield drop on CHI-B1

96.4% vs 98.7% baseline · 4h ago

link_off

Edge offline · PRG-C2

Last heartbeat 2h ago · check power

psychology

Model retrained · pcb-solder-v3 → v3.1

Recall +0.6pts on solder bridge · auto-deployed

12h

Defect mix · 7d

1,210 total · top 6 categories

Surface scratch41234%

Solder bridge28724%

Mold flash19816%

Low fill14212%

Crack988%

Other736%

Yield by site · 30 days

Hover any line to drill in

BangaloreChihuahuaPragueOsaka

dashboard

One pane of glass

Every line, every site, one screen. Yield, defects, throughput - live and side by side.

notifications_active

Alerts that find you

Yield drift, model regressions, offline edges. Slack, email, PagerDuty - pick your channel.

auto_graph

Drift detection

We watch the model watching your line. Confidence drops trigger an audit before throughput does.

group

Role-based access

QA leads see their line. Plant managers see the plant. The COO sees everything. SSO + SCIM included.

history

Audit trail · 365 days

Every verdict, every image, every model version - replayable. Designed to support ISO 9001 / IATF 16949 audits.

cloud_off

Optional, never required

Inference runs on the edge. The cloud is for visibility, not operation. Pull the plug - the line keeps going.

Minutes, not shifts

designed to detect yield drift before the next batch

Every verdict

replayable - image, model version, threshold

Live, not nightly

yield, defects, throughput across every line

Optional, never required

edge keeps inspecting if the cloud is gone

Pick the right level of cloud.

Move up when you need to. Air-gap if you have to.

Capability

No-Cloud

Lookout

See it on your part.

20 minutes. Bring a sample. Get a working detector.

AI visual inspection for production lines — Edge appliance + Cloud console.

Specifications, not slogans.

AI inspection that runs on your line.

Many defects. One clear call.

Tune, dry-run, plug in.

Inspection Studio

Simulator

Bring your own cameras

One small box.Sits next to your line.

Camera in. Verdict out. Under 50 ms.

Pick the rig that fits your line.

Every line, every site, one screen.

A console that turns inspection into decisions.

Fleet overview

Pick the right level of cloud.

See it on your part.

One small box.
Sits next to your line.