From supplier PDF to your PIM, automatically

What you are actually looking at

Real documents, not a staged dataset

Five supplier datasheets, including a German-language transmitter TDS, run through the same pipeline any upload goes through: extraction, audit, and structuring. On the battery datasheet in this recording, 24 of its 25 extracted fields landed as typed PIM attributes.

A script over the REST API, not a connector

The push is a short script calling the PIM’s standard REST API, recorded against a self-hosted Akeneo Community Edition instance. SpecMake does not ship a PIM connector and is not trying to replace your PIM. What SpecMake gives you is the structured, audited, source-anchored data, through the app or through our REST API, so your integration layer can post it wherever it needs to go.

Units and provenance survive the trip

Measurements arrive as typed attributes with the unit in the attribute rather than buried in a string, and every record keeps the source file, revision, detected domain, and language it came from. That is what makes the data auditable after it lands. More on how the pipeline gets there in how SpecMake works and the extraction feature page.

What happens, scene by scene

The complete caption text from the recording, in order.

1. 1

   Title card

   From supplier PDF to your PIM, automatically. A real workflow, recorded end to end on a live Akeneo CE instance.

2. 2

   The source document

   The input: a real supplier datasheet, a 48 V battery module, as a PDF.

3. 3

   Structured output

   SpecMake extracts and audits it into structured JSON: every value typed, with unit, confidence, and the exact source quote.

4. 4

   The push

   A short script pushes that JSON into the PIM through its standard REST API.

5. 5

   The catalog

   Five datasheets became four products and one product model, families assigned, 100% complete.

6. 6

   Typed attributes

   Values arrive typed: 51.20 Volt, 100 Ampere hour. Units live in the attribute, not in a string.

7. 7

   One document, three sizes

   One datasheet described three valve sizes side by side.

8. 8

   Product model and variants

   So it became a product model plus three variants: shared values live once on the parent, each variant carries only what differs.

9. 9

   Provenance

   Every record keeps its provenance: source file, revision, detected domain, and language.

10. 10

    Closing card

    The data preparation layer for your PIM. Extract, audit, structure, then push to the platform you already use.