Use case: technical descriptions from your manufacturers' datasheets

By Manuel Mattar4 min read

In short

In this example, the distributor stores the PDF datasheets next to each product, moves the key specifications into a file it imports as attributes and source data, sets up its brand's technical voice and generates by family. Today Bruxel does not read the PDFs when generating: it writes from the imported data.

This is an example scenario, not a customer. The quantities are hypothetical and help frame the problem.

The situation

An electrical supplies distributor sells about 3,000 products from 25 manufacturers: switches, contactors, cable, light fixtures. Each manufacturer sends its datasheets as PDFs, and everything that matters is there: voltage, current, protection rating, standards. In the store, by contrast, most products only have the ERP name and a code.

The team wants correct technical descriptions, with the tone of a distributor that knows what it sells, without inventing specifications.

First: what Bruxel does with a PDF today

Before designing the process it is worth being clear. Today Bruxel stores datasheets next to each product, private and virus-checked, but does not read them when generating. A datasheet in PDF, HTML or text starts marked “Use to describe the product”; that mark gets it ready for when generation reads datasheets, but it does not change what is generated yet.

What does feed generation is the product data and the source data: the columns of the file you import. So the work is in moving what matters from the datasheet into columns.

What the team does in Bruxel, step by step

1. Store the datasheets where they can be found

On each product page, Files tab, the team uploads the PDF with “Upload document”. Bruxel gives it the “Datasheet” type and the tab groups files by type. In Media → File types the team also creates a “Safety data sheet” type with an expiry date, so the tab says when it is about to expire. Datasheets are downloaded with a link that lasts 5 minutes and only after the virus check.

2. Move the specifications into a file

The team (or each manufacturer, if it already has them in a table) builds a file with one row per SKU and one column per specification: voltage, rated current, number of poles, IP rating, standard. It uploads it in Catalog → Import. In “Review before importing”, Bruxel says what it would do with each column:

  • if it matches an existing attribute, it fills in on its own, with no credits and no AI;
  • if it does not exist yet, it is created as a filterable attribute with its values;
  • if it is not useful as a filter (a long text, a different value per product), it stays as source data and keeps feeding generation.

Filling in from the file never overwrites a value someone fixed by hand, so you can re-import when a new datasheet arrives.

3. Set up the technical voice

In Content → Brand voice, the team picks the form of address, describes the tone (“precise, no filler adjectives”) and writes the context a copywriter of theirs would know: that it sells to electricians and installers. In forbidden words it lists the terms its legal team does not approve. By default Bruxel writes without efficacy claims or guaranteed results, and the team leaves it that way.

4. Generate by family

In Catalog → Products, it filters a family, for example contactors, and uses the panel to generate the descriptions of the products in that filter. “See how many and what it costs” gives the price before charging. In a batch it costs 7 credits per product instead of 10, and descriptions arrive within a few hours; they are tracked in Content → Batches. Only what comes out right is charged.

5. Review and correct

Bruxel checks that each description has no filler text, no broken HTML and is not in another language; if something fails, it is flagged and not published. When a description confuses a product (“this is not a contactor, it's a relay”), the team uses “Correct” on the product page and the correction stays with the product.

What gets solved

  • Datasheets live next to the product, sorted by type and with expiry dates.
  • Specifications become filterable attributes in the store, not just text inside a PDF.
  • Descriptions are written from the manufacturer's data and in the distributor's voice, at scale and with a known cost before launching.

What does not get solved

  • Bruxel does not read the PDFs when generating. Moving the specifications into a file is still the team's or the manufacturer's work.
  • Bruxel does not compare the description against the PDF. It checks text quality, not that every figure matches the datasheet; that review is human.
  • Web search does not replace the manufacturer's datasheet: it brings public information that may not match the exact model.
  • The voice is one for the whole catalog. If a product line needs another tone, today it is not configured per family.

To go deeper, read the guides on technical documents, generating descriptions and brand voice, or the Files, AI content and Import pages.

Frequently asked questions

Does Bruxel generate the description by reading the manufacturer's PDF?

Not today. A datasheet in PDF, HTML or text can be marked Use to describe the product, and that mark gets it ready for when Bruxel reads datasheets while generating, but it does not change what is generated yet. Generation uses the product data and the source data you import.

Is web search worth turning on for a technical catalog?

It helps when the product has little more than a name, and it adds 5 credits per product. If you already imported the datasheet specifications, it adds less. For technical products, your manufacturer's data is more reliable than what a third party publishes.

What do I do if a description comes out with a wrong technical fact?

On the product page, use Correct on the generated field and write what is wrong, the way you would tell a copywriter. The product is generated again with the correction, and the correction stays for future generations in all your languages.

Sources

  1. Datasheet (Wikipedia)
  2. Product information management (Wikipedia)

About the author

Manuel Mattar

Founder of Bruxel.AI

In 2014 he founded Wolf Sellers, an e-commerce agency in Mexico City, and later Bruxel.AI, the AI-powered PIM.

See all posts by Manuel Mattar

Try it with your catalog

Create your free account and try Bruxel with your own products using the welcome credits.

Keep reading

Use case: technical descriptions from your manufacturers' datasheets