Automotive
Weld seams, torque sequences, paint defects and line-side safety across body, paint and final assembly.
Inspect · GuardVisual Intelligence Infrastructure
AI that sees the physical world
VRTLST turns every camera on your floor into a sensor that understands what it is looking at — defects, hazards, people, process, throughput — in real time, at the edge.
person · 0.98
forklift · 0.94
pallet · 0.91
hard-hat · 0.88
VRTLST Vision
One model that understands every camera you own
Detection, segmentation, tracking and classification from a single deployment — no per-camera retraining.
- Works with the RTSP cameras already on the wall
- Fine-tune on 50 labelled frames, not 50,000
- Runs cloud, on-prem or fully offline
FAIL 2 defects · confidence 0.96 · diverted to rework
VRTLST Inspect
Catch the defect before it leaves the line
Sub-millimetre surface inspection at line speed, with a decision the PLC can act on.
- Pass / fail in under 50 ms, in-cycle
- Learns new defect classes from operator feedback
- Writes straight to MES, SCADA or the PLC
person in zone
09:41:02 Missing hi-vis · Dock 3 warning
09:38:55 Tailgate at Gate A warning
09:31:10 Fire lane obstructed cleared
VRTLST Guard
Every zone watched. Every rule enforced.
Draw a boundary once. Guard enforces it on every shift, on every camera, without a human staring at a wall of screens.
- PPE, exclusion zones, tailgating, fire-lane obstruction
- Hardware e-stop and access-control integrations
- Faces blurred on capture — evidence without surveillance
06:0010:0014:0018:0022:00
1,284units / hr
99.2%yield
4m 12sdowntime
VRTLST Watch
Know what's happening the moment it happens
Continuous monitoring across every line and every site, with alerts that reach the right person in seconds.
- Live throughput, cycle time, dwell and idle detection
- Anomaly alerts to Slack, Teams, SMS or PagerDuty
- Ninety days of searchable event history
INFERENCE38 ms
THROUGHPUT60 fps
POWER12 W PoE
RATINGIP67 · −20→60°C
VRTLST Edge
The camera that thinks for itself
A sealed edge-AI device that runs the full VRTLST stack on-board. Single PoE cable in. Decisions out.
- Full inference on-device — works with the network down
- Zero frames leave the plant unless you say so
- Fleet updates and rollback from Command
PLANT 1
PLANT 2
PLANT 3
DEPOT
LINE A
LINE B
DOCK 3
GATE A
PLANT 3 Inspect model drift · retrain suggested review
VRTLST Command
Every camera, every site, one screen
The control plane for your whole visual estate — deployments, model versions, alerts, access and audit.
- Roll a model to 200 cameras, or roll it back
- Cross-site benchmarking and drift detection
- SSO, role-based access and a full audit trail
Running on the floor at
NORTHRIDGEKAWA MOTORSSteelbrookOMNIPARTVanteraHELIOS FOODSARCLIGHTDensmark
0% defect recall in production
0 ms median edge inference
0M frames understood per day
0 frames leaving your network
The platform
Six products. One way of seeing.
Vision is the engine. Inspect, Guard and Watch are what it does for you. Edge is where it runs. Command is where you steer it. Nothing to stitch together.
Command
Control plane · fleet · access · audit
Inspect
Defect detection
Guard
Safety & security
Watch
Monitoring
Vision
Detection · segmentation · tracking · classification
Edge
On-device runtime · cameras · gateways · air-gapped sites
Under the hood
What happens between the lens and the line
Five stages, one model, thirty-eight milliseconds. Scroll to walk through a single frame.
- 01 · Capture
A frame arrives
RTSP, GigE, thermal or depth — 30 to 60 times a second, from a camera you already own.
- 02 · Detect
Objects are found
People, vehicles, parts, tools and PPE, each with a confidence the rest of the stack can reason about.
- 03 · Segment
Edges are resolved
Pixel-level masks separate the part from the fixture, so a measurement means something.
- 04 · Track
Identity persists
The same object keeps its id across frames — that is what turns detection into cycle time and dwell.
- 05 · Decide
Something happens
Pass, fail, divert, alert or e-stop. A decision the line can act on, inside the takt time.
person · 0.98
forklift · 0.94
pallet · 0.91
id 07 id 12
CLEAR no breach · cycle 41.2 s · logged
VRTLST Vision
The core computer-vision platform
One model stack that handles detection, segmentation, tracking and classification — trained on industrial footage, not internet photos. Point it at a camera and it starts working.
- Any camera
- RTSP, USB, GigE Vision, thermal and depth. Keep the hardware you have.
- Few-shot training
- Fine-tune a new class from 50 labelled frames in an afternoon.
- Your data stays yours
- Nothing is used to train shared models. Ever.
operator
torque tool
fastener ×6
VRTLST Inspect
Manufacturing defect detection
Scratches, porosity, misalignment, missing components, short fills, bad welds. Inspect calls it in-cycle and tells the line what to do about it.
- In-cycle decisions
- Pass, fail or divert in under 50 ms — inside the takt time.
- Operators teach it
- A one-tap correction on the HMI becomes tomorrow's training data.
- Traceable
- Every unit gets an image, a verdict and a reason, stored against its serial.
FAIL diverted to rework · 0.96
VRTLST Guard
Factory security & safety
The rules you already have — PPE on the floor, nobody inside the robot envelope, fire lanes clear — enforced continuously instead of audited quarterly.
- Zones you draw
- Sketch a boundary on the feed. Guard watches it on every shift.
- Acts, not just alerts
- Trigger an e-stop, lock a door, sound a local horn.
- Privacy by default
- Faces blurred at capture. No identity, no biometric matching.
person in zone
09:41 Missing hi-vis · Dock 3 warning
09:31 Fire lane obstructed cleared
Where it runs
Six industries, one model stack
The physics of a weld seam and a fill level are different. The way you teach a model about them is not.
Electronics
Component presence, solder quality and orientation at speeds no human inspector can hold for a full shift.
Inspect · VisionFood & beverage
Fill levels, seal integrity, foreign-body detection and hygiene-zone compliance, wash-down rated.
Inspect · WatchLogistics
Dock safety, load verification, pallet counts and trailer utilisation, from the yard to the pick face.
Guard · WatchHeavy industry
Exclusion zones around plant, slag and spill detection, and vision at sites with no cloud connection.
Guard · EdgePharmaceutical
Fill and finish inspection, gowning compliance, and an audit trail that satisfies the quality system.
Inspect · CommandNot on the list?
If it has a camera pointed at it, the answer is usually yes. Send us an hour of footage and we will tell you.
Talk to an engineerEverything else
The rest of the stack
Watch keeps an eye on the whole operation. Edge puts the model in the camera. Command ties all six together.
Watch
Real-time monitoring
Throughput, cycle time, dwell and downtime across every line and site — live, with alerts that find the right person.
Explore Watch Edge
The edge-AI camera
A sealed IP67 device running the full stack on-board. One PoE cable in, decisions out — and it keeps working when the network doesn't.
Explore Edge Command
The dashboard that brings everything together
Deployments, model versions, alerts, access and audit for every site — one control plane instead of six.
Explore CommandIntegrations
Fits the stack you already run
VRTLST is a sensor and a decision, not a replacement for your control system. It writes into what you have.
Control layer
- OPC UA
- Digital I/O
- Modbus TCP
- PROFINET
- EtherNet/IP
Plant systems
- MES
- SCADA
- Historian
- ERP
- CMMS
Data & messaging
- REST
- MQTT
- Webhooks
- S3 / Blob
- Kafka
People
- Slack
- Teams
- PagerDuty
- SMS
Customers
What it looks like on the floor
Automotive · Inspect
Weld-seam scrap cut by two thirds across four lines
Read case study →Logistics · Guard
Zero recordable dock incidents over three consecutive quarters
Read case study →Food & beverage · Watch
Nine hours of hidden micro-downtime found in week one
Read case study →Heavy industry · Edge
Vision at an air-gapped site with no cloud connection at all
Read case study →In their words
Do we have to replace our cameras?
No. Vision runs on standard RTSP, USB, GigE Vision, thermal and depth cameras. Most sites go live on the hardware already mounted. VRTLST Edge is there for new installs, harsh environments and sites where you want inference physically on the device.
Does our footage leave the building?
Only if you choose cloud deployment. On-prem and Edge deployments keep every frame inside your network — the air-gapped configuration has no outbound connection at all. Your footage is never used to train shared or third-party models under any deployment.
How much labelled data do we need to start?
Around fifty labelled frames per new defect class for a first working model. The base models are pre-trained on industrial footage, so you are fine-tuning rather than starting from zero. Operator corrections on the HMI feed the next training round automatically.
How does it talk to our line?
Digital I/O and OPC UA for PLC-level decisions, plus REST and MQTT for MES, SCADA and historians. Inspect can write a pass/fail verdict directly against a serial number. Alerts route to Slack, Teams, SMS, email or PagerDuty.
How long does deployment take?
A single-line pilot typically runs two to four weeks from kickoff to a model making live decisions. Multi-site rollouts are staged through Command, one line at a time, with rollback available on every deployment.
How is it priced?
Per camera stream, per year, with volume tiers by site. Edge devices are bought outright or leased. Pilots are fixed-fee and credited against the first year if you continue.