Stable decision
Agreed thresholds and one OK/NOK/Error logic instead of subjective judgement.
Visual inspection for mid-sized manufacturers
Visual inspection your own team can deploy.
AIVQC turns manual visual inspection into a repeatable OK/NOK decision at the line. You mark the defects and automatic training prepares the model — no machine-vision expert, no integrator costs and no images sent to a cloud.
For plants with a quality engineer and IT staff on site. Start with one station in a 30-day pilot — on your own or with our support.
- ✓ No machine-vision expert
- ✓ No integrator costs
- ✓ No cloud — works offline
- 01 Image capture
- 02 Local analysis
- 03 Decision and trace
Interface demonstration — not a reference validation result.
COMPLAINTSA repeatable decision instead of a tired eye
NO ONE TO INSPECTThe station checks every part on every shift
YOUR HARDWAREWe check the camera and computer you already have first
SMALL SITENo separate server — the Trainer computer is enough
More complete than a smart camera.
Less complex than enterprise machine vision.
WHY AIVQC
From manual inspection to a repeatable decision.
An operator can see a defect. AIVQC helps turn that knowledge into a measurable, versioned process for every captured part.
We do not sell “just a model” or integrator hours. We give your team tools that combine imaging, data, training, validation and line operation into one controlled process.
Every part, every shift
The same decision logic in the first and last hour of a shift — no fatigue and no differences between operators.
Data under control
Production images do not have to leave your infrastructure.
Full traceability
You know which model, recipe and configuration produced every result.
WHO IT IS FOR
For plants that want automation without a big project.
AIVQC works best in mid-sized manufacturing plants — where visual inspection is manual today and a classic machine-vision project is too expensive or too slow.
IT IS FOR YOU IF
- ✓You visually inspect a repeatable product — by hand today, at the line or at packing.
- ✓The defect can be shown clearly in an image: flash, a missing part, a scratch, dirt, a wrong label.
- ✓You have a quality or process engineer and IT staff on the team.
- ✓Production images must stay inside the plant.
- ✓You want to start with one station and see the result on your own line.
WHAT EACH ROLE GETS
Quality manager
A repeatable OK/NOK decision, measurable false accepts and false rejects, and a history of every inspection with its image.
How to set decision thresholds →Plant manager and management
A pilot on one station before investing, and deployment without integrator costs — with a calculation based on data from your own line.
How to calculate the return →IT department
No cloud and no ports open from the Internet, encrypted backups and automatic updates. At a small site, no separate server.
How AIVQC Server works →DEPLOYING WITH YOUR OWN TEAM
Have an engineer and IT staff? You can deploy it yourself.
We designed AIVQC to be launched by the plant's own technical team, not an external integrator. You pay for a software licence, not for a specialist's hours — and the knowledge of the system stays in the company.
- 01
Quality or process engineer
Marks the defects, sets the training goal and time, reviews the result. Automatic training chooses the model and parameters.
- 02
IT administrator
Installs Server as a Docker container or — at a small site — as a service on the Trainer computer. The installer guides each step.
- 03
Automation engineer
Connects the station to the PLC with a few Modbus TCP signals and tests everything in simulation mode.
- 04
Operators
Start inspection with one button and see the OK/NOK result with its decision time.
WHERE YOU SAVE
- +No integrator hours — for the first deployment and for every change.
- +A new product variant is work for your engineer, not an order and a lead time with a supplier.
- +You use the camera and computer you already have, if they pass qualification.
- +Additional lines repeat the same path with the same people.
Missing an automation engineer or unsure about the image? We can help in one area or run the pilot together with you.
AIVQC PILOT
Try AIVQC on your line — with us or on your own.
The pilot is the full platform on one station. You train a model on images of your own product and see the result on your line before deciding on deployment.
- PILOT SCOPE
- TRAINER
- 1 engineering workstation
- PRODUCTION
- 1 inspection station
- DEFECTS
- up to 3 defect classes
- TIME
- 30 days + a free extension on request
After the pilot
A production licence and deployment — with us or your own team. With an in-house deployment you pay for the licence, not for integrator hours. We quote individually.
Pilots and deployments are currently available in Poland only.APPLICATIONS
Defects visible to a human.
Decisions made by the system.
A good image is essential for a good inspection. The best first application is a repeatable product, controlled position and a defect that can be shown clearly in an image.
PLASTIC HOUSING
Flash detected
NOKASSEMBLY
Component missing
NOKPRODUCT LABEL
Position correct
OKExample process scenarios — not reference validation results.
01↗02↗03↗04↗05↗06↗
Missing component
Check assembly completeness and the presence of required components.
Incorrect assembly
Detect reversed, displaced or incorrectly seated parts.
Flash and deformation
Identify moulding, geometry and product-edge defects.
Cracks and damage
Visually detect scratches, cracks and surface damage.
Labels and print
Verify the presence, orientation and legibility of markings.
Contamination
Detect foreign objects, dirt and surface deviations.
HOW AN INSPECTION GOES LIVE
AI deployed in a controlled way.
A model is not the end of the project. Only an approved package, station qualification and unambiguous decision logic create a production solution.
- 01COLLECT
Collect images
Natural process variation and real OK and NOK examples.
- 02TRAIN
Prepare the model
Annotation and automatic training that chooses the model and parameters itself.
- 03QUALIFY
Set criteria
Quality, errors and performance measured on an independent test.
- 04DEPLOY
Publish the package
The verified model goes only to selected stations.
- 05IMPROVE
Improve the process
History, difficult cases, a new version and safe rollback.
HOW THE PLATFORM IS BUILT
From the first image to the decision on the line.
Three specialised components prepare, publish and run an approved inspection.
01TRAIN
AIVQC Trainer
Mark the defects — automatic training prepares and checks the model, no machine-vision expert needed.
- +Automatic training
- +Leak-free data
- +Evaluation like the line
02CONTROL
AIVQC Server
Publish approved versions only to selected stations.
- +Local deployment
- +Access control
- +Release history
03INSPECT
AIVQC Production
Make local OK/NOK decisions and pass them to the process.
- +Local OK/NOK decision
- +Fail-safe fault handling
- +Modbus TCP and history
RETROFIT, NOT REVOLUTION
We check the hardware you already have first.
AIVQC qualifies the computer, camera and configuration for a specific process instead of requiring one closed hardware ecosystem.
- 01 Camera, optics and lighting assessment
- 02 Decision-time measurement on the target computer
- 03 Replacement recommendation only where necessary
DESIGNED FOR PRODUCTION
The line does not wait for the Internet.
AIVQC Production keeps the latest verified package locally. Inspection can continue during a server or network interruption, while activation and rollback remain controlled.
- 01 Local copy of verified packages
- 02 Permissions matched to the user's role
- 03 The system detects loss of inspection capability
- 04 Inspection history in a local database
PLANT NETWORK / ON-PREMISES
TRAINEREngineering workstationAPPROVED PACKAGE
SERVERLocal Docker / NASVERIFIED DELIVERY
PRODUCTIONEdge PC · Line 01 READY FOR OFFLINE OPERATION
EVIDENCE, NOT A PROMISE
Every inspection application receives its own validation result.
We do not transfer a demonstration result to your process. We measure the solution on representative data and under target imaging conditions.
QUALIFICATION REPORTGO / CONDITIONAL GO / NO-GO
- Product and conditions
- product type · camera · lighting
- Data
- image count · defect classes · test split
- Quality
- recall · precision · false accepts and rejects
- Performance
- decision time · throughput · target hardware
A reference validation result is not a guarantee for other products or process conditions.
AIVQC VS OTHER APPROACHES
A productised deployment process without being locked into one camera.
We compare categories of solutions, not specific products. Capabilities depend on the vendor and the project — treat the table as a list of questions to ask each of them.
| Capability | Smart camera | All-in-one AI kit | Cloud platform | Integrator project | AIVQC |
|---|---|---|---|---|---|
| Automated OK/NOK decision | ✓ | ✓ | ✓ | ✓ | ✓ |
| AI-based defect detection | partly | ✓ | ✓ | ✓ | ✓ |
| Training without a machine-vision expert | depends | ✓ | ✓ | — | ✓ |
| Your own defect classes and decision thresholds | depends | depends | ✓ | ✓ | ✓ |
| Inspection works without Internet access | ✓ | ✓ | depends | ✓ | ✓ |
| Images and training stay in the plant | ✓ | depends | — | depends | ✓ |
| Reuse of the camera and computer you already have | — | — | depends | depends | ✓ |
| Controlled publishing of versions to selected stations | depends | depends | ✓ | depends | ✓ |
| Role-based permission to change thresholds | depends | depends | depends | depends | ✓ |
| No fees per number of images | ✓ | ✓ | usually not | ✓ | ✓ |
| Deployment by your own team, without an integrator | depends | ✓ | ✓ | — | ✓ |
| Pilot on your own line before you buy | depends | depends | depends | depends | ✓ |
We describe each approach and six questions worth asking vendors in Smart camera or AI system? Four approaches to visual inspection.
FOUNDER-LEDINDUSTRIAL AI
WHY AIVQC EXISTS
I built the tool I needed on the production floor myself.
I worked in a manufacturing plant where visual inspection was manual — repetitive, tiring and dependent on who happened to be at the line. Automating it meant an integrator project that was hard to find budget and time for. AIVQC grew out of that need: a system the plant team can deploy, understand and improve on its own — without a cloud and without machine-vision expertise.
Dawid Oleśko · Founder, AIVQCFAQ
Frequently asked questions.
Every station and process is different. A pilot replaces assumptions with a real test.
See all questions (62) →What size of plant is AIVQC for?+
It works best in mid-sized manufacturing plants starting with one to a few stations. In offline mode, Server runs on the Trainer computer and serves up to three stations, so you do not need a separate server. Larger plants install Server as a Docker container and add more lines.
What does the pilot involve?+
The pilot is the full platform on one station: one Trainer workstation, one inspection station and up to three defect classes for 30 days, with a free extension available. You train a model on images of your own product and see the result on your line before you decide on deployment. A pilot with our support ends with a report recommending GO, CONDITIONAL GO or NO-GO. Step by step: AI quality-inspection pilot.
Can we deploy AIVQC ourselves?+
Yes. If you have your own process engineers and IT staff, you have everything you need: Server installs like any Docker container — or, at a small site, as a service on the Trainer computer — and automatic training needs no machine-vision expert. You pay for a software licence, not for integrator hours — usually the biggest saving. Contact us for a quote.
Who do we need on the team to deploy AIVQC ourselves?+
A quality or process engineer who knows the product and its defects, IT staff to install Server, and an automation engineer to connect the station to the PLC. None of them needs to be a machine-learning specialist. Details are in the guide Deploying with your own team.
Can I use my existing camera or computer?+
Yes, if they pass qualification with the target camera, model, recipe, I/O and required cycle time. A processor or camera name alone does not guarantee the result. How to assess it is covered in Retrofit or new station?.
Do my images have to go to the cloud?+
No. Training runs on your own computers, and image analysis runs at the AIVQC Production station. Inspection and package publishing work on the plant network, also without Internet access. More in Why production images do not have to leave the plant.
What if AIVQC goes out of business?+
Your line keeps running. We commit that in that case AIVQC will keep receiving security updates for at least 2 years and licences will be made perpetual. Images, datasets, models and inspection history are on your computers, and models are stored in the open ONNX format.
How much does AIVQC cost?+
We quote individually — a pilot and deployment with us, or the software alone if your own team deploys it. You do not pay for a cloud or for the number of images processed, because training and inspection run on your hardware. How to work out the return using data from your own line is covered in How to calculate the return on automating quality inspection.
CONTACT
Let's talk about your process.
Describe the product, defect and current inspection method. We will make an initial assessment of whether the application is suitable for a pilot — and whether your own team can deploy it.
contact@aivqc.com- No-obligation response
- Data sent directly to AIVQC
- Initial inspection-application qualification