AI in Base: Automatic Product Listings and Translations

Base – one of the most popular multichannel sales management tools in Poland – has announced a series of AI-based features. It's not a single function but a whole line-up of tools, the first of which is automatic completion of product listings: names, descriptions, photos and translations. In this article we show exactly what has changed, what real value it brings to online sellers and where – judging by the announcements – Base is heading next.

Prefer to listen?

If you'd rather listen than read – watch the episode of our podcast Biznes w Internecie (in Polish), where we discuss what's new in Base and share our own experience from deployments for multichannel clients.

What exactly did Base announce?

The new features work directly in the Products module – no exporting data to external tools. Specifically, AI in Base:

• generates the product name from data already saved in the listing – especially useful after importing a wholesaler's catalog, where names tend to be chaotic (e.g. "PROD_2847_XL_BLACK");
• writes the product description, composing it from the product information that's already available;
• removes photo backgrounds with one click – just hover over the thumbnail and click the scissors icon, no Photoshop or external graphics tools (supported formats: JPG, PNG, GIF, WEBP, up to 2 MB);
• translates content using DeepL, in the same fields where you edit names and descriptions – no exporting or juggling files between tools.

All of these actions can be run one by one for a single product or in bulk for the whole catalog. In the same series Base also announced Autosell – AI Listing Agents meant to list up to 100 products on a marketplace in a single run, without filling in data by hand.

What value does this bring to online sellers?

For stores managing large product catalogs, it answers three problems that show up in practically every online store:

• Time. Manually filling in hundreds of product listings is one of the most time-consuming and least loved jobs in e-commerce – repetitive, monotonous and error-prone at scale. Where it used to be measured in days, it's now measured in minutes.
• Distributor photos. A classic problem for multichannel sellers: the distributor or manufacturer supplies one photo on a white background, sometimes poor quality, sometimes none at all – and a white background is required on practically every marketplace. Automatic background removal solves this without involving a graphic designer.
• Translations. Entering a foreign market – whether Amazon or another marketplace abroad – usually means weeks of translating the entire catalog. Here it happens in the same fields of the product listing, with no content exports. While you're at it, it's worth making sure translated descriptions don't drift apart between sales channels – inconsistent product content makes catalog management harder in the long run.

One caveat is worth stating plainly, because Base itself emphasizes it: the AI composes the description from the available product information – the better the listing is filled in, the better the result. This isn't a tool that creates content out of nothing – if a product listing has two sentences and nothing else, the AI will generate two nicely formatted sentences, not a ready-made sales text. In practice, though, most stores already have some set of data from the distributor or manufacturer, so it's more of a starting point for enrichment than a real obstacle.

In practice, the biggest difference shows up in bulk work. Until now, refreshing the description, title and photos for a whole catalog – for instance after switching suppliers or during a periodic content update – meant going product by product, often manually across several places at once (the panel, an export sheet, a graphics tool). Now the same thing can be selected as a group operation across the entire catalog at once. For a store with several hundred or a few thousand SKUs, that's the difference between a project spread over weeks and a task you can finish in a single afternoon.

Worth knowing – labeling AI-generated images (the AI Act)

Since August 2, 2026 the EU's AI Act (Article 50(4)) has been in force, requiring content that meets the definition of a deepfake to be labeled – that is, images or videos generated or substantially modified by AI so that they look real but aren't. Allegro is already informing sellers of this obligation and preparing a tool for automatically labeling such images. In that context, will photos modified in Base get such a label too?

Simply removing the background of a product photo most likely doesn't qualify as this kind of manipulation – it's an edit, not the generation of a new scene from scratch. The obligation applies where a product photo is, for example, placed into a fully generated setting. If Base moves toward generating complete product photos in the future (rather than just removing backgrounds), that's a different category – and there an "AI" watermark will be mandatory, regardless of whether the photo was prepared by the seller, an agency or a wholesaler.

What other AI features will Base offer?

Base presents its new features in the "AI in Base" series on its blog. Upcoming features already announced include:

• Autosell – AI Listing Agents that list products on marketplaces automatically;
• AI Assistant in the panel – an assistant for working with product data and account actions;
• AI-based product counting in the WMS (Warehouse Management System);
• AI-dons – extensions written with the help of AI;
• ACP (Agentic Commerce Protocol) – a store available directly inside ChatGPT;
• MCP (Model Context Protocol) – Base data available in the AI tool of your choice.

The last two points – ACP and MCP – fit the broader direction we already covered when writing about agentic AI in e-commerce: the store as a data layer that AI agents talk to directly, without the classic interface in between. It's the same trend we're seeing across the market – and Base seems to be taking a concrete step in that direction, not just making an announcement.

It's worth adding that there's a solid financial foundation behind this: in 2025 Base reported revenue of PLN 300 million and profit of PLN 159.5 million – profitability above 50% at that scale. It shows that the investment in AI isn't an experiment with a question mark over it, but profits being reinvested in the development of a tool already used by a large share of the Polish e-commerce market.

Summary

Base is clearly investing in AI features and consistently delivering the announced series – a new feature every two weeks is a pace we haven't seen from this tool before. Automatic product listing completion, photo background removal and DeepL-based translations are changes many sellers managing large catalogs have been waiting for – especially those selling across multiple channels and abroad. The upcoming features, from Autosell to AI tool integrations (ACP, MCP), show that Base increasingly thinks of itself as a full product management system, not just a sales synchronization tool.

If you use Base and want to check whether your product listings are ready for these changes – or you need help with automating sales processes and listing offers on marketplaces – write to us. We'll gladly help you figure out how to make the most of these features in your store.