What's new in Nucleo

Stock has one owner: Commerce

Stock quantities now live in Commerce only. The Catalog shows them read only, on the product page and in the variants table, with a link to Commerce › Inventory where they change, and the Inventory entry is gone from the Catalog menu.

Imports, the Catalog API and Atomo set stock only as the initial quantity of a new variant. Afterwards a stock line is skipped with a note (imports) or refused with a clear message (Catalog API), so the numbers you see everywhere are the ones Commerce reads from Shopify, or writes there when Nucleo runs your orders.

In the Catalog chat, Atomo's changes wait for your click on Confirm

When Atomo shows a preview of a change in the Catalog chat, only your click on Confirm under the preview applies it. Writing "yes" in the chat is no longer enough, and Atomo can no longer confirm a change on its own, even if a text it has read asks it to.

Product titles, descriptions and attribute values can come from imports, Shopify or suppliers, so Atomo now treats them as untrusted text. Once it has read them in a conversation, every change goes through the preview, including a small edit right after opening a product. In voice mode a spoken yes still confirms, unless Atomo has read untrusted text: then you confirm with the button in the chat.

Parent categories include their subcategories' products

A parent category now counts and finds the products of its subcategories. The category tree shows both numbers, the category's own products and the total with its subcategories, and filtering search by a parent returns everything underneath it.

In the category editor, the products that come from subcategories are listed apart and read only, so you always know where to change them. Merchandising order stays set category by category.

Atomo reorganises your category tree in one step

Ask Atomo to restructure your categories and it moves them all in a single operation: everything moves or nothing does, and search is updated once at the end. Before acting it sums up the change in a couple of sentences, how many categories move and where, instead of listing every one, so you can say yes in a glance.

A home that tells you what to do next

The Catalog and Commerce homes are rebuilt around Atomo. A thin strip at the top shows the date, one chip per connection (Shopify, your PLM, the Catalog API, the Nucleo engines) with the detail on hover, and a single number for where you stand. Below it, a greeting and the Atomo box: type, dictate or talk, and your recent conversations are one click away.

Under the box, Next: one line per thing worth your attention, in a fixed layout with the count, the priority and a single action, Open or Fix with Atomo. Set aside what you do not care about and it stays out. Further down, what Atomo did overnight, what it is doing now and what waits for your OK, plus the setup guide when you are getting started.

Shopify setup: Nucleo as your product source, in five steps

The Shopify setup is now five steps: Setup, Products, Collections, Customers, Orders. Markets, languages, locations and every flow's direction live in the first step, each step has one action, and the data is already there when you open it.

You choose once who manages product information. With Nucleo in charge, it writes everything it has mapped to Shopify (texts, options, variants, photos, tags, metafields, collections) and leaves what it does not know untouched, including other apps' tags; a change made by hand on Shopify is put back. With another system in charge, Nucleo only reads.

Every import brings in the whole catalog. Metafield definitions, key:value tags such as Gender_Men and product options are proposed as attributes to confirm, matching the ones you already have instead of creating duplicates.

Stock has its own update setting (Real-time, Every night or Manual), switching a flow's direction no longer waits on thousands of products, and flow settings read From Shopify, To Shopify, Both ways or Disabled.

Shipping methods you set up, and no more sample values

Commerce now has a Shipping methods page: you list the methods you offer, with a price in each currency, starting from the ones already used on your Shopify orders. When customer service changes an order's shipping, only those methods are offered.

Screens across Catalog and Commerce no longer fill gaps with sample data or a guessed country, currency, language or time zone. Countries, currencies, languages, store hours, holidays and return reasons come from your settings, and when something is missing the page says so and links to where you set it.

Changes you make in Nucleo are no longer undone by Shopify

Editing a product, a smart collection's rule, an order's shipping method, cancelled quantities or a released hold in Nucleo now holds even when Shopify sends its next update. Products scheduled to go live no longer turn into drafts in Nucleo.

Smart categories now count everywhere: in product feeds, search filters and the Catalog API. Draft categories stay out of the storefront, and the Catalog API shows the price of a running sale. Editing a location, attribute set or synonym only changes the fields you touched.

Shopify follows your sales, channel texts and category order

When Nucleo writes to Shopify, a sale on your default price list now reaches the storefront with the full price struck through, and a market price list in draft stops pricing that market. The texts, SEO and translations you set for the Shopify channel, and the attributes you mapped to metafields, go out with each product.

Publishing a category's order in merchandising sorts the matching Shopify collection. Product feeds and the Catalog API count stock only from active locations that ship orders, and scheduled channel publications are respected by the API too.

Publishing and sales scheduled to the minute

A product's publication window on a channel and a sale price list's validity now take a date and a time, not just a day. On Shopify the product goes live, or back to draft, within a minute of the time you chose, and a sale's prices start and stop the same way. Dates set before keep their meaning: a sale that ran until a day still runs until the end of that day.

Shopify appears as a sales channel only once your store is connected, and new workspaces start without sample sales channels.

Choose who updates whom between Nucleo and Shopify

Every Shopify flow now has a direction: Shopify updates Nucleo, Nucleo updates Shopify, both ways, or ignored. Products, collections, prices by market and stock offer all four; orders always come from Shopify, and both ways lets Nucleo send fulfillments, cancellations and refunds back. Catalog flows also choose how often they update: automatically, every night, or only when you ask.

The source of a flow controls it fully, deletions included, and switching a flow to write to Shopify asks for confirmation first. Products go to Shopify complete, with variants, options and photos, and only when they're published on the Shopify channel. A completed store stock count is always sent to Shopify.

Bring your whole Shopify store into Nucleo

A new setup wizard brings what's already in your Shopify store into Nucleo, in five guided steps: markets, field mapping, products, collections and prices per market, then stock, customers and order history. Preview your markets before applying them, choose which products come in with include and exclude rules, try the import on 20 products, then import everything with progress and resume.

Tags and metafields get a mapping proposal you confirm or correct, and it keeps applying to every later update. A Setup and flows card on the Shopify connection shows each flow's status, last update, records and errors, with one-click realign. In Commerce, Record only lets orders come in as a record while order management isn't live yet. Nucleo only reads your store: nothing is written to Shopify.

Product cost on the product page

The product cost now has its own field in the Info tab, under Compliance & customs: enter it once and it is saved on every variant. When one size costs more or less, change it in the new Cost column of the Variants tab.

Catalog API: one season, one channel, one language

The Catalog API can now download just the products of one or more seasons, and just the products published on a sales channel with the texts that channel shows, in the language you need.

Each sales channel can also have its own translations in every language — title, description, translatable attributes and SEO — on top of the product's own: your online store and a marketplace can show different German texts. Edit them on the product page by choosing the channel and the language, or read and write them through the API. Option names, like colors, can already differ by channel in the attribute settings, and the API now returns them too.

Simpler data from connected systems: whoever can write, writes

Per-field ownership rules are gone. A connected source, such as a PLM or an ERP, writes the fields it sends, and the last write wins. Nothing is locked in the product editor: anyone who can edit a product — and Atomo, imports and the API — can change any field.

Access is now just read or write: give a system a read-only Catalog API key if it should never change your catalog. Every change shows up in the product's history with its author, and can be restored. Each connection shows what the source sent and the fields the catalog doesn't know.

For existing Catalog API integrations: keys that were set to never overwrite ("Nucleo" rule) are now read-only; keys set to fill only empty fields ("insert" rule) now overwrite existing values; `GET /ping` no longer returns `key.field_rule`; the `409 field_rule` and `409 field_locked` errors no longer exist.

Deleted products go to the Trash for 30 days

Deleting a product now moves it to the Trash instead of removing it for good. It disappears from the catalog, search, feeds and exports, and comes back with its variants, prices, images, categories and history whenever you restore it.

The Trash shows who deleted each product and when it will be removed for good, after 30 days.

Saved views that follow you and can be shared with your team

Saved views of your lists are now stored in Nucleo, not in your browser: they follow you from one computer to another, and you can share a view with your whole team. Views you had already saved were moved over automatically.

A setup guide for new companies in Catalog and Commerce

New companies get a setup guide on the Catalog and Commerce home pages: connect your store, import products, add photos, invite your team. Steps tick themselves off as you go, and the guide disappears once everything is done — or whenever you hide it.

Asset comments, activity and sharing saved for the whole team

Comments, activity and team sharing on assets are now saved and visible to everyone on your team, signed with the real author, and the permission you choose when sharing — view, comment or edit — is enforced.

Custom roles: decide exactly what each person can do

Organization › Roles lets a company admin build roles that fit the way their team works. Start from a built-in role or from scratch, then pick permissions section by section — edit products but not prices, work orders without touching shipping settings, build dashboards without managing connections. Catalog, Commerce and Intelligence all follow the same roles, and so does Atomo: it can never do more than the person it works for.

Built-in roles keep working exactly as before and can be duplicated as a starting point. Assign a role from a person's page or right in the invite, and changes take effect within seconds — including when you take a permission away.

Atomo builds you a workspace, right next to the chat

Ask Atomo to show you something — “show me the black t-shirts that sold best this month” — and instead of a list in the chat it opens a workspace beside the conversation: product grids, tables, key numbers, charts, side-by-side comparisons, with real data read live from your catalog. The page underneath stays where it was; close the workspace and you're back.

Pin the views you want to keep, reopen them from Atomo's history or as a full page, and export the products in any view as a complete CSV. When Atomo prepares a file in the chat, a download button appears right in the conversation.

Changes that need your go-ahead now show up as preview cards — what will change, before and after — and Confirm applies exactly what you saw.

Hand Atomo your files — and teach it how you work

Drag, paste or attach photos, PDFs, CSV and Excel files, and text straight into the Atomo panel. Atomo reads them, works on spreadsheets, and quotes its sources: under each answer you see the sentence and page of the PDF, or the website, it came from. Give it a supplier file and it prepares the products to create in an editable grid, flags the problems, and creates them as drafts once you're happy.

Atomo can also search the web for a product's missing details. What it finds never goes straight into your catalog: it arrives in Enrichment as a suggestion, with a link to the page it came from, for you to approve.

Teach it once and it remembers. Say “remember that…” or “from now on…” and Atomo keeps it for the whole company; save a sequence of steps as a procedure and reuse it whenever you need. Both are listed in Catalog › Configuration › Atomo, and memories picked up in chat apply once an owner or admin approves them. Overnight, Atomo works through the filter attributes with the most gaps, so fresh suggestions are waiting in the morning.

Findability: can shoppers find what you have in stock?

Search › Findability measures how much of your available stock your customers can actually find. Press Measure now and Nucleo replays the real searches made on your store, then shows the share of available products that appear in the first two pages, the best sellers no search finds, and the searches that return nothing — most asked first — with the history of every measurement.

Then let Atomo fix it. “Make available products findable” starts a piece of work that groups its proposals with the evidence behind each one, lets you decide group by group, applies what you approve and measures again on the same searches, so you see the before and after.

Claude and ChatGPT can now reach Catalog, Commerce and Intelligence

The Nucleo connector for AI assistants such as Claude and ChatGPT now covers every module. Alongside Brain, an assistant connected with your Nucleo login can work with your catalog, read customer care tickets and requests, store locations and stock in Commerce, and pull sales overviews, KPIs, trends, top products and channel reports from Intelligence — always in your language and never beyond your permissions.

Small, easily reversible Catalog changes go through right away, as they do in Nucleo. Anything that would need your confirmation in Atomo — deletions, bulk edits, imports, publishing — is shown as a preview and confirmed only inside Nucleo.

Product versions, only when you want them

Editing a product never creates a version: you correct and enrich it as always. When a product genuinely changes from a certain date — a new composition, a revised fit — use the arrow next to Save and choose Save as new version…, with the date it's valid from and an optional name like “rev. B”. Imports can do the same with the Create versions from this file switch, reading each row's Valid from date or Version.

Products with versions get a Version menu under their publication status. Open any past version to see the product as it was, read-only, with the earlier value noted under each field that differs; compare two versions to see only what changed; rename a version, move its date or delete it. Channels always receive the current version, and the Product versions export gives every version with its start and end dates, ready to cross with your sales.

Data ownership, simplified

Who writes each field is now decided in one place only: the rules of each connection. For every group of fields, and down to a single field or language, you choose Source updates — the source always keeps it current — Empty fields only — the source fills a field only while it's empty in Nucleo — or Nucleo owns it, and the source never writes it.

With the rules doing the work, there's no longer a separate list of differences to review: what your rules allow is applied, and what they don't is left alone. Every change still shows up in the product's history, with restore.

Atomo moves to the latest AI models

Atomo now runs on Claude Sonnet 5.5 and Claude Opus 5.5 across chat, enrichment and batch work, for sharper answers and better reasoning on long tasks. The OpenAI and Gemini options are refreshed too, with GPT-6 Sol and GPT-6 Luna, Gemini 3.8 Flash for heavier work, Gemini 3.5 Flash-Lite for quick tasks and Gemini 3.1 Flash Image for images.

A more polished Catalog, everywhere you look

We went through Catalog screen by screen. Pages load with skeletons that match what's coming, never with a misleading “empty” or “0”; every deletion and bulk change asks first, and its message arrives only once the result is in. Form errors appear under the field that caused them, required fields are marked, and the Back button, tab switches and the product arrows protect your unsaved work — the save bar simply shakes to remind you.

Every table now shares the product list's layout: search with the magnifier, the columns button on the right, a sticky header and the same pagination. On the product list you can select all matching products, not just the page, undo a status change, and use keyboard shortcuts — / to search, j and k to move, ⌘S to save. Product photos always show whole, never cropped.

Catalog speaks your language end to end: thousands of interface texts, including the messages that come from the server, now read naturally in English and Italian. Categories can now have their own image, price lists and sale lists can be deleted, redirects can be edited, and the price grid shows each product's photo.

Catalog fixes

The product list search no longer drops letters as you type. Saving or publishing in Merchandising always goes to the channel you're working on. Category members now include every product, not just the first page, and long sale lists load page by page. Saving a product no longer gets stuck, numbers like 1,522 display correctly, and people with view-only access no longer see buttons they can't use.

Every barcode of a variant, each with its type

The Variants table now has a Barcode column with each variant's primary code and how many more it has. Edit barcodes lists every code of the variant with its type — EAN, UPC, GTIN, ISBN, ASIN, Custom or your company's own code types — each checked against its format. The first one is the primary.

Code types are managed in Settings › Catalog data. Imports rename a SKU only when the file says so, with a previous SKU column, and stop any row whose barcode or SKU already belongs to another product, so codes are never silently mixed up.

Discovery shows what Nucleo adds to each product

The Discovery tab of every product now holds only what Nucleo adds on top of what you entered. What Atomo read in the photos comes first, followed by a single Recommendations block: alternatives and pairings side by side, the widget that answered with a link to its editor, the top suggestions, and this product's own exceptions. Try in playground opens the same product and logic in the Search Playground.

Pick products by hand and only those appear, in your order — the widget adds nothing. What you fill in yourself moves to where it belongs: SEO to the bottom of the Info tab, where Product type and Brand now sit side by side, and the product's pinned positions to the Categories tab.

One save bar, and a product page that works like Shopify's

Discard and Save have left the page header. As soon as you change something, an Unsaved changes bar appears at the bottom of the screen — on products, attributes, price lists, widgets, categories, Merchandising and Configuration — and it only offers Save when there's something that can be saved.

The product page header gathers Copy link, Duplicate and Delete under More actions, with arrows to move to the previous or next product of the list you came from, filters and order included. Page titles now match the menu, so you always know where you are.

Recommendations, rebuilt around widgets

Recommendations now has its own place in the Catalog menu. Every block of suggested products is a widget: when you create it you choose whether it shows Alternatives — similar items to consider instead — or Pairings — items that go with the one on screen. Each widget has its own editor page and a live preview on any product, so you see the result before you save.

A widget can cover the whole catalog, a category, a filtered set or a single product, and the most specific one always wins. Conditions compare candidates with the product being viewed using AND/OR logic — same line, same color, a different category — and pinned products stay exactly where you put them. On the product page you can see which widget answered, and why.

Merchandising is reorganized around it: Sorting for the order of products in each category, Global rules for pins that apply across every category, and Strategies.

Atomo gets on with it — and shows you what it did

When a request is clear, Atomo now simply does it: creating a product or an attribute, updating a handful of items, filling in suggested values. It stops to ask only where a mistake would cost you — deletions, changes to many products at once, imports and exports, publishing to a channel, activating a price list, and in Brain sending, merging or signing.

When it's done, the page next to the conversation opens on what just changed: create an attribute and its page appears, update a product and you're looking at it. The same rule applies in Brain and in Catalog.

Product activity as a clean timeline

The Activity tab on every product is now a compact timeline — one line for who changed what, and when, whether it was a person, an import, a connection or Atomo. Expand any entry to see exactly which values changed and restore the previous one in a click.

Atomo can now work on your whole catalog

Ask Atomo the way you'd ask a colleague: “select all the black t-shirts and set them to draft”, “create a 20% sale price list for the summer collection”, “add a Fit attribute and translate it”. Atomo builds the selection from any attribute, in any language — variant values included — and applies the same bulk actions you'd use by hand, with every change recorded in the product's activity under its name.

Beyond bulk edits, Atomo reaches everything Catalog does — attributes, categories, translations, price lists, channel views and the media library — always within your permissions. New sale price lists are created switched off, so nothing reaches customers until you decide.

A more natural voice for Atomo — now in Brain too

Atomo's voice now runs on a full-duplex model: it listens while it speaks, lets you interrupt, and sounds like a native speaker of your language. Every spoken request goes to the same Atomo that powers the chat, with the same tools and permissions — whatever you can do by typing, you can do by talking.

Brain gets the same voice experience as Catalog, and its Atomo chat now shares the look of the Atomo panel.

Use your company's own AI provider keys

Owners and admins can now add their company's own Anthropic, OpenAI or Gemini API keys for Atomo, in both Brain and Catalog. When a key is set, Atomo's chat, voice, enrichment and batch work for that provider run on your account; leave it empty and Nucleo's keys take over, so nothing ever stops working.

Atomo conversations are now kept per module and per company: Catalog chats stay in Catalog, Brain chats in Brain, and switching company switches history. Your existing conversations came along.

Product filters that scale to dozens of attributes

The search-and-filter bar on the product list is built for catalogs with dozens of filterable attributes: search fields by name, browse them in groups — Main, Data quality, Attributes — and see how many products match each value before you pick it. Active filters show as clear two-part pills, with the total number of results right next to the bar.

The table header stays pinned while you scroll the page, and the filter bar comes back as soon as you scroll up. Fixed along the way: “is not” filters now exclude what they should, and filters on multi-select attributes return the right products.

Decide which source owns each field — and see every change

For every connection that brings data into Catalog, you now decide who owns each group of fields, with exceptions down to a single field or language: the source keeps it updated, Nucleo keeps it, or the source sets the first value and Nucleo takes over from there. Nothing is preset — the rules are yours.

Connecting a source never overwrites your work: differences are listed for you to resolve, and so are later updates that clash with a field Nucleo owns. An empty value from a source never wipes yours, and fields the catalog doesn't know yet stay visible until someone decides what to do with them. Every product also keeps a full change history — who changed what, when and from which source — with restore.

Talk to Atomo

Atomo now listens and answers out loud. Tap the microphone in the Atomo panel and have a real conversation about your catalog: ask a question, create a product, fill in its details. Spoken answers are short, Atomo asks one thing at a time, and it starts talking as soon as the first sentence is ready.

Choose the language you're speaking right next to the microphone — English, Italian, French, German or Spanish; it starts from your interface language and is remembered. Atomo can also apply a product's suggested values for you: you see the preview, confirm once, and it's done.

Klaviyo joins the connector library

Klaviyo is now in the Catalog connector library: connect it with your API key and Nucleo creates the matching channel for your product data, or adopts the one you already had.

Channels across the app now carry their official brand marks — Klaviyo, Google Merchant Center, Cettire and Tmall among them — so destinations are recognizable at a glance.

A cleaner media library, with real video previews

Videos in the media library now show their first frame as a poster, in both grid and list view, with a play mark — instead of a colored placeholder. The same poster appears on the asset page's player.

The library itself follows the Catalog layout: folders and trash in a rail on the right, the same command bar and tables used across the module, and linked products that show their cover and title and open straight to the product page.

A new Catalog home: what happened while you were away

The Catalog home opens with a greeting, today's key numbers one click away from their pages, and “While you were away”: what the catalog did on its own in the last 24 hours — orders read from your store, new proposals from Atomo, alerts, merchandising signals refreshed, customer profiles and search index rebuilt.

Atomo's queue is grouped by urgency — do now, today, when you can — and can be filtered by where each finding comes from: catalog data, product photos, orders or searches. Every card says what would change, with the action, a hand-off to Atomo, or dismiss. A Connections panel shows each sales channel and Nucleo service with its last run, status and numbers.

Merchandising that knows what sells — and what comes back

Category sorting can now weigh signals refreshed every night from your store: momentum over the last 30 and 90 days, returns — leaving out size exchanges, which say nothing about the product — the share of sales made at a discount, repurchase rate and carryover across seasons. A stock guardrail sends items with less than two weeks of cover to the bottom, so you never promote what's about to run out. Signals only count once an item has sold enough to be meaningful.

Pick the channel you're working on from “View as” in the top bar: merchandising and facets follow it everywhere.

A noticeably faster Catalog

Product lists, product pages and the Catalog home now load much faster. We removed work that ran on every page without being used, cut the round-trips behind the list and the product page, compressed what travels to your browser and prepare the heaviest summaries right after each release — so even the first visit of the day is quick.

Catalogue profile: search that adapts to any catalog

Search, recommendations and size advice no longer assume fixed attribute names. They ask for roles — which attribute is the color, the size, the gender, the season, the product line — and the new Catalogue profile maps each role to your own attributes, whatever you sell.

Atomo proposes the mapping with the evidence behind each choice, and a person signs it off. Until a role is signed, the features that depend on it stay quiet instead of guessing.

Enrichment becomes a focused work queue

Enrichment now sits under Catalog, next to Products and Inventory, and works one step at a time: pick an attribute and it opens full width with Atomo's proposed values, each on its own card with Approve and Reject — and the name of whoever decided.

Searches that found nothing, together with the product copy Atomo suggests to answer them, now live in Search → Queries, which links straight to the right fix. Old addresses redirect.

Storefront signals: what shoppers do on your site

A lightweight storefront snippet now sends Catalog what shoppers do on your site — searches, product views, add-to-carts and purchases — only after they've given consent. Events feed search insights, merchandising and customer profiles, and a customer is recognized only when your store vouches for them, never from an event alone.

In Shoppers, the Signal tab shows what's arriving and from where, and Test events lets you check an installation before it goes live.

Shoppers: what your customers teach your catalog

A new Shoppers section builds a profile for every customer from their order history: what they buy, their usual size, how much they spend and how often at a discount, the language they order in and the behavior groups they belong to. Look anyone up by order number, see a whole group at a glance in Everyone, save the groups that matter and export them.

The part that pays off is what shoppers teach the catalog. Return reasons written by customers, items that run large or small according to those who exchanged sizes, and searches with no answer become work queues on the attributes to fix — each with its own remedy.

Returns on every product, and a fairer completeness score

The product list has a new Returns column — the return rate, with the units behind it — and you can sort by it without an item sold three times jumping to the top. Each product page also shows the searches customers used to find it.

Completeness is now stricter and consistent: the list and the product page score the same way, and products no longer earn points for fields their attribute set doesn't declare.

Complementary products: what goes with this item

Beyond similar products, Catalog now suggests the ones that go with the item a shopper is looking at. Pairs are learned from what customers actually buy together, a category map says which categories complete each other, and merchandisers approve suggested pairs in groups or write their own by hand.

Each item counts once, whatever its colorways, and suggestions respect the shopper's size when it's known with confidence.

Personalized results, and a Playground to see them

Search and recommendations can now take the shopper into account: their usual size, worked out from the purchases they kept, and how each item fits, learned from customers who exchanged sizes — “runs large”, “runs small”. When a size isn't certain it's suggested, never forced.

In the Playground you choose who is searching and compare the same results with and without their history, with an explanation of why each result moved.

Search Insights on a single screen

Search Insights is now one dashboard with a period selector, the way you know it from your analytics tools: the real searches from your storefront, top terms, searches that led to a click and those that found nothing, by language and by source. Search settings went from seven tabs to three.

Know where every attribute value comes from

Each attribute on a product page now shows where its value comes from — an import, a person, the product description or the product photos — and which values are still unknown. Atomo's suggestions also start as soon as you leave the description field.

Site search, built on your catalog

Catalog now includes its own site search. It reads your catalog as it is — attributes, variants, prices, translations — and understands what shoppers type: typos, plurals, colors, sizes and negations such as “without hood”, turned into filters on your real attribute values in every store language.

Merchandisers manage synonyms and redirects, choose facets and ranking, and try everything in a Playground before it reaches the storefront. Every query is logged, so Insights shows what people look for and what they can't find.

Atomo reads your product photos

Atomo now looks at product images and reads what they show — colors first of all — then checks them against your catalog data, flagging where the photo and the record disagree so you know which one to fix. Fields the text doesn't state are deduced where possible, and the queue raises only what's genuinely wrong.

Search uses it too: colors seen in the photos answer queries like “black and green”, and product descriptions now count when matching.

Catalog health, attribute suggestions and inventory metrics

The Catalog home now opens on the health of your catalog: closed seasons still online, sold-out items, missing attributes, duplicate options and empty SEO fields — each with a count and a ready-made filter that opens exactly those products. Anything you don't need to act on can be set aside.

Atomo proposes attribute values read from product titles and descriptions, choosing only among the options your catalog already uses and in the language you ask for. Inventory gains sell-through, days out of stock, remaining cover and sizes that never sold.

Multilingual SEO for every product

Product SEO now speaks every language your store does. Meta title, meta description and URL handle can be tuned per language, so each locale ships copy written for it instead of one shared version stretched across markets.

Switch language right inside the SEO tab: the title, brand and attribute tokens resolve in that locale, and any field you leave untouched falls back to your SEO rule-set pattern for that language. Editing one language never affects another — each locale is stored on its own.

Product type, standard on every product

Every product now carries a dedicated Product Type field across all stores — the same taxonomy your sales channels expect, ready to map straight through when you publish. Set it per product or fill it in bulk, and it travels with brand and the rest of your attributes.

Bulk-assign brand, product type and category

Select any number of products in the catalog and set their brand, product type or category in one move. Categories add to what a product already has instead of replacing it, so you can layer classifications across a selection without losing the existing ones.

Export the products you're looking at

Filter the catalog down to exactly the products you want, then export that selection to CSV straight from the list — same format as the full export, so it round-trips cleanly back through import. No more exporting everything and trimming the file afterwards.

Nucleo Feed: perfect product feeds, images customized per channel

Nucleo Feed is live: one catalog in, validated feeds out — Google Shopping, Meta, marketplaces — with rule-based attribute mapping, per-channel and per-country exclusions, error validation that catches problems before the channel does, and scheduled exports that keep every destination in sync.

The standout: a built-in image generator that applies layers, backgrounds, badges and promo overlays to product images per channel, with the whole feed re-rendered automatically. Channel-perfect visuals without a design round-trip.

CSV data export

Every dataset in the PIM can now be downloaded as CSV — products, stock, markets, locations, currencies, languages, sizes, price lists, attributes and categories — straight from System Configuration.

Exports mirror the import layout, so you can pull a file out, edit it in a spreadsheet and load it straight back in without remapping a single column. Products come out one row per variant, with stock in its own export you can re-import to update inventory.

Categories in the product import

The product import now assigns categories from the file itself: add a Category column and each product lands in the right place in your catalog tree. Hierarchical paths like “Clothing > Tops” are created on the fly, and re-importing never duplicates an assignment.

Atomo lands on the PIM catalog

The Atomo AI assistant is now wired to the PIM: ask about products, attributes, completeness or channels in plain language and it answers from your live catalog, straight from the shell panel.

CSV product import

Import products from any CSV with interactive column mapping — bring catalogs in from spreadsheets or legacy tools without an integration project.

Completeness score and Promotions

Every product now carries a completeness score visible on the catalog list, so content gaps are obvious at a glance. A new Promotions system manages promo campaigns with full CRUD and bulk actions.

DAM live in the unified shell

The DAM module now reads live data inside the Nucleo shell — libraries, folders and assets come straight from the DAM backend, with image renditions proxied on demand.

Per-channel publishing

Products now have a publishing state per sales channel, with sync status, scheduling windows and market/POS scoping — publish to Shopify today and to a marketplace next Monday, from the same catalog.

Locations, channel pricing and SKU patterns

New Locations registry (address, capacity), per-channel price overrides on top of market price lists, and configurable variant SKU patterns so generated SKUs follow your own convention.

Category tree with channel overrides

A full category subsystem: hierarchical tree, per-channel overrides and one-click seeding of categories from an existing product attribute.

Per-channel product overrides

Override fields, attributes and tags of a product for a specific sales channel — one master record, channel-perfect content everywhere.

Bulk catalog actions

Set status or delete across hundreds of products in one action, plus per-store capability toggles.

Multi-country markets and price lists

Markets group countries with their own currency and per-country VAT; price lists live under markets, and attributes can be imported from an existing external PIM export.