Method
Structure first. Then repeatable work.
Industrial product information does not become reliable because it has been copied into a new format. It becomes useful when its source, structure, review, and intended market use are clear.
- 01
Source material
Input PDFs, spreadsheets, images, supplier data, and domain knowledge.
Work Collect the available material and identify what it is, where it came from, and what remains uncertain.
Output A visible source set rather than an invented product library.
Typical mistake Treating a supplier catalogue or spreadsheet as complete, consistent, and ready for publication.
People, software, and AI People judge relevance and source quality. Software records and organizes the material. AI can assist extraction where the source is clear enough to inspect.
- 02
Product model
Input The source set and the product questions the market actually needs answered.
Work Define families, variants, attributes, terminology, inheritance, and relationships.
Output A product model that makes the structure explicit and reusable.
Typical mistake Creating a flat table that loses the distinctions between a family, a variant, an option, and a related product.
People, software, and AI People define business and technical meaning. Software keeps the model consistent. AI can help classify material against an approved structure.
- 03
Source-linked knowledge
Input The product model and evidence from the original material.
Work Link claims to provenance, review them, record approvals, and make known gaps visible.
Output A knowledge layer that explains what is supported, approved, and still needs attention.
Typical mistake Hiding uncertainty to make a product record look complete.
People, software, and AI People own review and approval. Software preserves traceability and status. AI may flag inconsistencies, but it is not the source of truth.
- 04
Market assets
Input Reviewed, source-linked product knowledge and a clear channel purpose.
Work Adapt the same foundation into web-ready product libraries, catalogues, product cards, presentations, distributor packs, and localized materials.
Output Consistent assets for the people who market, sell, distribute, and implement the product.
Typical mistake Rewriting the same product story independently for every channel until the versions drift apart.
People, software, and AI People decide audience, priority, and approval. Software creates repeatable templates. AI can transform approved information into a defined format.
- 05
Repeatable operation
Input A proven workflow, its rules, and defined approval points.
Work Turn repeatable work into a software-supported operation while keeping human responsibility where a decision matters.
Output A system that can be improved, reused, and audited instead of rebuilt from scratch each time.
Typical mistake Automating before the source, model, and approval path are understood.
People, software, and AI AI where useful. Software where repeatable. Humans where responsibility matters.
From supplier material to a usable product library
Imagine a manufacturer with a supplier catalogue in PDF, a spreadsheet of product codes, a folder of images, and years of terminology held in sales and technical teams. The first task is not to publish a new catalogue. It is to identify the sources and gaps, build a product model that distinguishes families and variants, and review the meaning of the attributes.
Once that foundation exists, the same approved knowledge can support a web library, a product card, a catalogue section, and a distributor presentation. The source still matters; the human review still matters.
What the method does not claim
- It does not create truth from missing or contradictory source material.
- It does not replace domain review or commercial judgement.
- It does not turn a weak source into an automatically approved output.
- It does not treat AI as an independent authority.
AI is not the product. The product is faster, cleaner and more repeatable industrial work.