Automated Optical Inspection to Auto Cognitive Inspection

1. What is outdated in AOI is not the A, nor the I — it is the O

AI AOI already says the important part: a visual foundation model replaces hand-written rules. The letter that no longer fits is the one in the middle — O, optical. That single letter binds three things at once, and all three have stopped holding.

It binds the imaging modality

“Optical” defaults to visible-light imaging. Yet the same DaoAI World world-model foundation also runs structured-light 3D, X-ray, infrared and quantitative analysis of electron-microscope images. The decision logic is shared; the way the image is formed is not. Keeping the word “optical” writes the other modalities out of the category — and on a real line they usually coexist.

It binds the inspected object

In industry usage, AOI is close to a synonym for circuit-board inspection. When the same system moves into medical consumables, food packaging or chemical film, the word arrives with the wrong context: the first reaction is “we do not make boards”. Losing on the name before the solution is even read is the least useful way to lose.

It binds the decision principle

Optical inspection carries an implicit promise: take an image and compare it with a reference, pixel by pixel. This generation does the opposite — it recognises components, not pixels. The difference is not accuracy but what the criterion hangs on. Hang it on pixel differences and a new light source or a new incoming batch forces a retune; hang it on feature cognition and the question becomes “what should this component look like”.

2. C replaces O: Auto Cognitive Inspection

Three bindings, side by side

Bound by the “O”	What optical inspection implies	What cognitive inspection actually does
Imaging modality	Visible light by default; other modalities fall outside the word	Structured-light 3D, X-ray, infrared and electron-microscope images share one decision logic
Inspected object	In industry usage, effectively means circuit boards	One system spans electronics, automotive, semiconductor, battery, pharma, food, chemical and consumer scenarios
Decision principle	Capture an image, compare it with a reference pixel by pixel	Recognise the component, not the pixels: the criterion hangs on normal component appearance

ACI (Auto Cognitive Inspection) keeps the A for automated and the I for inspection, and swaps the O for a C. The name asserts one thing only: the decision basis is cognition, not the imaging method.

Three things change with it, and each is already documented on existing product pages — none of this is a new promise:

How models are built — from collecting a defect library to learning from good samples only: the model learns normal appearance instead of enumerating faults.
How changeover works — from reprogramming board by board to zero-code changeover: one good sample completes automatic programming.
How decisions are made — from “flag anything over a pixel threshold” to semantic false-call filtering: telling real defects apart from normal variation that merely looks like one.

Taken together, “optical” is no longer the common factor among them. Cognition is.

3. Why change only one letter

Category creation usually dies from one thing: requiring customers to learn a new vocabulary.

In practice, tender documents, process specifications, inspection procedures and acceptance criteria all say AOI. An entirely new word means rewriting those documents — and the people who rewrite them are rarely the people who buy the equipment. That friction is larger than any technical evaluation.

Leaving A and I untouched and replacing only the O lets the body of the existing paperwork carry over: in most places AOI is an equipment category name or a process position, and renaming it does not change what those clauses mean. Moving the category forward while keeping edits to existing documents as small as possible is the cheapest path available.

4. Which generation do you need

You can check this without adopting the new term. If any of the following holds, the problem you are solving has already moved outside what “optical” covers:

Your line runs two or more imaging methods (visible light plus 3D, or X-ray, infrared, electron microscopy) and you want them to share one decision logic.
The inspected object is not a circuit board, yet every vendor description you read is written in PCB terms.
Changeover is frequent and each one requires reprogramming or a fresh defect dataset.
False calls come not from poor capture but from normal variation judged as a defect (reflection, texture, in-tolerance displacement).

If none of the four holds, an existing AI AOI deployment is already sufficient and there is no need to migrate — that is our reading too.

Frequently asked questions

Are ACI and AI AOI two different products?

No. ACI is a more accurate name for the same technical line, not a separate product line. The existing 2D, 3D and software offerings remain the same systems; pages and model names do not change.

Do AI AOI lines already in production need to be rebuilt?

Not for the sake of a new name. Whether to upgrade depends on the four self-checks on this page, not on what the category is called — and on whether the spend pays back, which is worked through in AOI machine cost and ROI.

Why not simply call it “AI inspection” or coin something entirely new?

Because tender documents and process specifications say AOI. Changing a single letter lets the body of those tender and process documents carry over — most occurrences are a find-and-replace; an entirely unfamiliar term shifts the cost of rewriting the whole document onto the customer.

From AI AOI to ACI