Skip to content
    AI parameter extraction

    AI finds parameters in text and in photos

    From a product's name, description and images it pulls out the parameter (attribute) values that someone would otherwise retype into a spreadsheet by hand. It saves the result in the catalogue's uniform structure.

    Extraction and normalisation

    Raw data turns into usable filters

    AI does not work with text alone. It knows the parameter definitions, their type and unit, and whether they belong to the product or to a variant.

    Values hidden in the text

    AI reads the product's name and description and fills in the values of the parameters defined in its template.

    Read the guide

    Information from photos

    If a property is visible in an image, AI can use it as another source when filling in parameters.

    Read the guide

    Name mapping

    It suggests how to match the different field names in supplier feeds to your own parameters.

    Read the guide

    Consistent values

    It normalises different spellings of the same value into a single list of values, so the filters in your shop stay clean.

    Read the guide
    Supplier comparison

    AI pairs the parameters of two suppliers

    When two suppliers send the same properties under different names — Farbe and Colour, Gewicht and Weight — AI compares both the names and sample values and proposes a pair for each parameter, with a confidence level. When a pair fits, you create it with one click.

    • You confirm proposals while AI works on the next ones — the rows never move under your hands
    • You tick all high-confidence pairs at once and see their list before anything is created
    • The values move at the next sync, or right away from the data already downloaded
    Workflow

    Fill in parameters in three steps

    The template defines what we look for; the product data provides the context, and the results are saved to the product.

    1. 01

      You assign a template

      For a group of products, you define the required and optional parameters, including their types and units.

    2. 02

      AI reads the sources

      It processes the name, description and photos and looks for values that match the defined parameters.

    3. 03

      You check the quality

      You can edit the values and compare accuracy between models, and against later human corrections.

    You stay in control

    AI proposes. A person confirms.

    When you fill in values in bulk, you can see what has changed in the catalogue. We measure extraction quality against known values and also track the share of data that a person later overwrote.

    Read the guide

    Try extraction on your own products

    Bring a sample of your feed or catalogue and we'll look at which parameters AI can fill in without manual retyping.