Extraction from documents
Attributes read from PDFs, technical data sheets, catalogues, labels and spreadsheets.
AI extraction and cleansing of product data.
Data from PDFs, data sheets and catalogues turned into checked attributes, with the source kept.
When it makes sense
Your specifications live in PDFs, not in the PIM.
AI reads documents fast. The real work is checking what it read, so we build the check in from the start.
- Specifications are retyped by hand from PDFs and data sheets.
- The same attribute appears in different units and formats across sources.
- Duplicates and gaps show up only after products go online.
- Products are classified to ETIM or ECLASS by hand, one by one.
- Old catalogues and spreadsheets wait for a migration nobody has time for.
Product data in numbers
10M+ SKUs in our clients' systems.
Figures from our own projects.
- 10M+SKUs in our clients' systems
- 70+PIM and e-commerce implementations
- 300+projects delivered
- 30+experts and consultants
- 2010operating since
- 35+PIM implementations
- openProd in the Pimcore Storeour AI product for supplier PDFs and spreadsheets
- Pimcore Platinum Partnerone of two Platinum partners in Poland
Sources: LemonMind, November 2026
See it on a data sheet
From a PDF data sheet to a product.
openProd, our own product, reads a one-page data sheet, asks for six decisions and saves the product.
From a PDF datasheet to a product
Six decisions, one save, every value traced to its page.
Which documents should AI read first?
30 minutes, no obligation. Show us a few real data sheets and we tell you what can be read and checked.
How it works
From a document to checked data in your PIM.
Five steps, and people decide at the one that matters.
Document in
PDF data sheets, catalogues, labels and spreadsheets, as they are.
Values read
Each value is linked to its row, column or page in the source.
Matched to your model
Attributes, units, dictionaries and classes of your own data model.
Your team decides
Only uncertain values wait for a decision, with the source one click away.
Saved in the PIM
Approved data goes to Pimcore, Akeneo or Ergonode, with history and undo.
Scope
What we extract and clean.
We start from your documents and your data model, not from a template.
Units and formats
Units converted and formats unified to the rules of your data model.
Duplicates and gaps
Duplicate products and missing values found and listed for a decision.
Classification
Products mapped to categories and to ETIM or ECLASS classes, with rules your team can check.
Source trail
Every value links back to the row, column or page it came from.
Team approval
Changes are visible before saving, and each one can be undone.
How we work
Sample, pilot, rollout.
- 01
Sample review
You send a few real documents. We check what can be read and which attributes matter.
ResultAttribute list and a measure of success - 02
Pilot on one product group
We run extraction and cleansing on one group or one catalogue. Your team reviews the output.
ResultChecked data and a rollout decision - 03
Rollout
We extend it to more documents and connect it to your PIM, with rules for review.
ResultA repeatable process your team runs
Case studies
The data work behind it, in practice.
Classification and catalogue migration in Pimcore: the ground that AI extraction builds on.
ETIM / BMEcat mapping
GTV
A central Pimcore system extended with ETIM classification, BMEcat export and the ETIM Mapper tool.
Read case study
PIM, DAM and CMS
Ventia
Pimcore PIM, DAM and CMS for an HVAC distributor in Poland: automated product migration, brand pages, a knowledge centre and B2B leads.
Read case studyFAQ
Before we read your first document.
What documents can you extract data from?
PDF data sheets and catalogues, spreadsheets with merged cells or shifted columns, labels, images and e-mail text. We start with a sample of your own documents.
Is this the same as openProd?
No. openProd, our own product at openprod.io, imports supplier files into Pimcore, Akeneo or Ergonode. This service uses it where it fits, or a solution built for your systems, and also cleans data already in your PIM.
Does AI change our data without review?
No. Uncertain values wait for a decision, changes are visible before saving, and each one can be undone through version history.
Can you classify products to ETIM or ECLASS?
Yes. We map products to categories and to ETIM or ECLASS classes, with rules your team can check. More on ETIM and BMEcat.
What about files that suppliers send every week?
That is supplier onboarding: see openProd and the supplier portal. This page is about documents and data you already have.
How do we start?
Send us a few real documents. We check what can be read, agree the attributes and run a pilot on one product group. Background: how to clean messy product data.
Olgierd MrozikCEO, LemonMind
Start with one document
Send us one data sheet.
In a free 30-minute call we look at your documents and agree a pilot on one product group.
30 minutes. Google Calendar opens in a new tab.
- 2010 founded
- 300+ projects
- 70+ PIM and e-commerce implementations