Changelog

What we shipped and when.

Product updates for openProd, newest first. Each entry says what changed and where you will see it.

September 2026

  1. Finish a stopped analysis without paying for a new one

    • Import

    If an analysis stops after the AI has already delivered its result, you can now finish it from the stored result instead of starting a second paid analysis. Technical stops no longer tell you to change your source files, and a separate notice explains time limits. Save to catalog also asks for a confirmation before anything is written.

  2. Product counts on the family filter

    • Catalogue

    The family filter on the products list now shows how many products each family holds, is searchable, and lists families that contain products first. Counts match what the list shows, without variants.

  3. Fewer, clearer unit conversion questions in Data review

    • Import

    Data review asks one question per value that needs converting, and once per attribute instead of once per product. Conversions that follow from your earlier answers are accepted automatically, a value written as read is never asked about, and "accept all recommended conversions" closes the rest in one step. Earlier, on 10 Sep, you could already choose an explicit target unit in Data review and convert values end to end.

  4. Link attribute type

    • Catalogue
    • Import
    • MCP

    A new attribute type stores web links, each with an optional title. Links can be edited on the product card, shown in the Canvas, exported and restored from Version history. Data import recommends this type for columns that contain links, and the MCP server describes link attributes and returns them in a parsed form.

  5. Table attribute type

    • Catalogue
    • Import
    • MCP

    A new attribute type stores a small table with typed columns, for values such as a specification with several rows. You define the columns in the attribute drawer, edit rows on the product card (they stack as cards on narrow screens), and see a summary in the Canvas. Data import can save a compound cell from a source file as table rows; Data review asks once per column how to save it.

  6. Import very large files in parts

    • Import

    A workbook too large for a single analysis is now analysed and saved in parts by rows, and a variant family too heavy for one save is written in consecutive parts. A plain card shows the state of an import saved in parts, and a warning appears before a paid analysis of a very large source.

  7. One drawer for a value the dictionary does not have

    • Import

    When a value is missing from a dictionary attribute, a single drawer lets you use an existing option, add the new option, or skip the value. If the dictionary gained the option after the analysis, it is offered under "Use existing option". Since 13 Sep a missing option no longer makes the whole plan fail.

  8. Split a mixed import into families by category column

    • Import
    • Catalogue

    When a source file mixes different kinds of products, Data review can propose one product family per value of the category column and create each family with its own structure. You can move products between families from a product list in the family decision. Data review can also suggest a name for a new family and ranks the existing families that fit best (23 Sep).

  9. Decide conflicting values: keep one, save as separate attributes, or skip

    • Import

    When two values compete for the same attribute, Data review shows each value with its file, page and language and lets you keep one, save them as separate attributes, or skip them for that product. Groups of same-shaped conflicts can be decided once.

  10. Bulk decisions and relationship columns in Data review

    • Import

    Proposed attributes can be decided in one bulk window, grouped by kind with a clear recommendation per row and a sticky action bar. A column that links products to other products is decided once as a link, as text, or skipped, instead of once per value.

  11. Show in source for every extracted value

    • Import

    Next to an extracted value in Data review, Show in source opens the original file at the place the value came from. PDFs open at the cited page, spreadsheets show the sheet and cell in a wide preview, and the quoted text is displayed. You can check a value without leaving the decision.

  12. Undo last decision and clear all decisions

    • Import

    After you Decide on a value, a notice offers Undo, and Data review has an Undo last decision action. If someone else saved a decision in the meantime, you are told and the list is refreshed. You can also clear all decisions of an import after a confirmation.

  13. Data import steps named for what they do, with cost and time estimate

    • Import

    The Data import steps were renamed to describe what they do with your data, sources load in three tabs, and analysis stays locked until sources exist. The Sources step shows an estimate of credits and time before you confirm a paid analysis, based on similar earlier imports, and the analysis result opens with a summary of what is waiting for you.

  14. Data review as one decision queue with Decide on each row

    • Import

    Data review became a single queue of decisions, each row with a Decide button and what has been decided so far, and a drawer that opens directly on the decision. Headers show the goal of the step and credits on every step.

  15. Preview of what will be saved, product by product

    • Import

    Before Save to catalog, the preview lists products first, each opening onto a grid of its values, and the result lists what was saved for each product. Relationship proposals appear under Relationships with what will be written.

  16. AI credit usage in the top bar

    • Platform
    • Import
    • Assistant

    The top bar shows the AI credits you have used this month, and an organization total where your role allows it. Data import, the Assistant and connectors all count into the same figure, which refreshes after each operation. This builds on the shared credit counter introduced on 9 Jul.

  17. Assistant and Version history together in one product drawer

    • Assistant
    • Catalogue

    On a product, the Assistant and Version history share one tabbed drawer, so you can ask a question and review earlier versions without losing your place.

  18. Searchable list of imports and sources from the asset library

    • Import
    • Assets

    The list of imports can be searched and filtered, and your filters are remembered. A source file for Data import can be chosen from the asset library, and since 28 Sep several files from the library can be combined with files from disk.

August 2026

  1. A workspace for reviewing import decisions

    • Import

    Review gained a workspace that groups open decisions into kinds of work, opens every decision in one shared drawer, and shows attributes globally with one row per attribute. Transformations and conflicts are readable as searchable lists, and each product's whole review opens from its row.

  2. Save to catalog, including imports in several pages

    • Import

    An analysed import can be saved to the catalogue by a user with permission, and the result shows what landed. An import in several pages is approved once and executed segment by segment with honest progress (17 Aug), and matched products are updated instead of duplicated (12 Aug). Products no longer need a SKU to exist (13 Aug).

July 2026

  1. Relations between products as attributes

    • Catalogue

    Product relations can now be defined as attributes, so a relationship has its own name and rules and can be edited on the product card. Import understands relationship columns and compiles them into these relations.

  2. Data import analyses several files as one job

    • Import

    A single import job can hold several source files, including PDFs read visually, and keeps a history of past jobs. The analysis is independently checked before you see the result.

  3. Dashboard with catalog pulse

    • Platform
    • Catalogue

    The dashboard opens on a workbench with recently changed products, recent imports and system activity, a Catalog pulse, and a shortcut to create a product. The product Canvas also shows rows in a hierarchy, with variants under their main product.

  4. Pimcore connector with product sync control

    • Integrations
    • Assistant

    The Pimcore connector lets you choose products to synchronize, shows the state of each (not synced, needs attention, synced), and retries the ones that need attention. A product is marked synced only after the result is verified in Pimcore.

  5. Assistant on the product page

    • Assistant

    A product page has an Assistant panel that summarizes the product, points out gaps and proposes texts. Proposed attribute changes appear as cards you approve before anything is saved, and fields changed by the Assistant are marked.

June 2026

  1. Assistant inside Data import

    • Assistant
    • Import

    A persistent Assistant panel in Data import answers questions about the current import and can apply actions you approve, such as accepting mappings or dismissing a validation issue. It asks structured questions when it needs your input and shows where its evidence comes from.

  2. AI-assisted data import with mapping review and validation

    • Import

    A new Data import flow reads your files, proposes how they map to your catalogue, lets you review the mapping, validates the result and then imports products, with estimates of AI usage. Variants and update-existing strategies are supported.

  3. Names translated per language across the catalogue

    • Catalogue
    • Categories

    Names of units, attribute groups, attributes, categories, families and dictionaries can be entered in every catalogue language. The drawers show language tabs, and required fields are marked with a suffix.

May 2026

  1. Audit log

    • Platform
    • Security

    An audit log records who changed what and when across units, attributes, families, products, assets, categories, settings, users, roles and sign-in events. It can be browsed as a chronological list, with retention settings.

  2. Version history for assets

    • Assets

    Every change to an asset creates a version you can browse, compare and restore, and you can upload a new version of a file while keeping its public link. Old versions stay previewable and retention is configurable.

  3. Categories

    • Categories
    • MCP

    Categories add several parallel trees to classify products, with ordering and product assignments and a Categories tab on the product page. Category changes are recorded in Version history, and the MCP server can read and change categories.

  4. Sign-in, invitations, users and roles

    • Security
    • Platform

    Users sign in with a login screen, are invited by email and can reset their password. Administrators manage users and roles under Settings, and the interface hides actions a user is not permitted to perform.

  5. Image, gallery, video and file attributes

    • Catalogue
    • Assets

    Products can hold an image, a gallery, a video or one or several files as attribute values, chosen from the asset library.

  6. Canvas export and assets library

    • Catalogue
    • Assets

    The product Canvas exports to CSV or XLSX. The assets library offers upload, folders with breadcrumbs, a resizable sidebar, thumbnails and previews for images, PDFs and video.

  7. MCP server for your catalogue

    • MCP

    An MCP server lets AI tools connect to your catalogue over a secured endpoint. It can look up units, dictionaries, attributes, families and products, and with scoped write access it can create and update them.

See it working on your own files.

Book a short demo and we will run a real supplier file through openProd with you.

Book a demo Explore the product