AI in logistics and e-commerce: why it all starts with a WMS

At the recent “Logistics of the Future” panel at the Transport and Logistics of Croatia conference, the question was simple: what does AI really mean for logistics? The answer our founder Matija Fontana gave there was just as simple: AI is only as useful as the data it works with.

In this article we show how AI, a WMS, warehousing and e-commerce fit together, where AI already helps today and how to get started without major investment.

AI without data is just an expensive conversation

AI tools can already write an email, summarise a document or analyse a spreadsheet. In logistics, though, the questions are concrete: how much stock do we have, where are the goods, which orders are late, which item is returned most often and why.

AI cannot answer those questions if the data does not exist or is scattered across emails, Excel sheets and paper delivery notes. That is why the first step is always the same: data has to be accurate, up to date and in one place.

Or, as Matija put it on the panel: “How will we develop larger infrastructure when we don’t have the data on which to base its development?”

The WMS is the foundation

A WMS (Warehouse Management System) runs everything that happens in the warehouse: goods receiving, locations, stock, orders, picking, packing, dispatch and returns. Every one of those steps leaves a trace in the data.

That data is exactly the raw material AI needs. Without a WMS, a warehouse has goods. With a WMS, it also has a history: what came in, what went out, how long it took and where it got stuck.

If you want to know more about the system itself, read The future of inventory management with advanced WMS solutions and What is a WMS and does your webshop need one.

Where AI already helps in logistics today

We are not talking about robots driving themselves around the warehouse. We are talking about applications that can be introduced step by step, on top of the data you already have:

• Stock planning: order history can be used to estimate when an item will run out and when to reorder, including seasons and campaigns.
• Warehouse slotting: the most frequently ordered items can sit closer to the packing zone, which shortens picking routes.
• Order batching: orders with similar items can be grouped into a single picking round.
• Pre-dispatch checks: invalid or incomplete addresses and unusual orders can be flagged before the parcel goes to the courier, instead of coming back.
• Customer replies: “where is my parcel” questions can be answered automatically from shipment status data.
• Documents: data from delivery notes, invoices and purchase orders can be entered into the system automatically, with a human check.
• Returns: return reasons can be categorised to show which products, packaging or webshop descriptions are causing problems.

They all have one thing in common: they only work if the WMS and the webshop already record what is happening. AI doesn’t replace people; it replaces chaos.

What this means for webshops and B2B

For a webshop, the key is that orders, stock and shipments are connected. Integration with Shopify, WooCommerce, Magento or directly via API means an order reaches the warehouse on its own, and the stock level flows back to the webshop.

For B2B and distribution the data is different but just as important: pallets, delivery windows for retail chains, returnable packaging, campaigns and kitting. The same rule applies: without accurate data there is no good planning and no good automation.

Dependence and cost: a question rarely asked

The panel also raised a question that gets less attention. Many of the AI tools we use every day rely on a handful of large American technology companies. “We are currently paying an incredibly low price for their services,” said Matija, warning that this could change quickly.

What this means in practice:

• Know which data you give to which tool, especially customer and pricing data.
• Your data should stay in your own systems, not only in the tool.
• Don’t build processes that stop if one tool raises its prices or changes its terms.

Small can move faster: 5 steps to get started

Croatia is a small market, and the large-scale robotics seen around the world is still too expensive for most companies. But small and mid-sized businesses can start right away:

1. Map where your data lives today: webshop, ERP, Excel, emails.
2. Connect the key systems: orders, stock and shipments in one place.
3. Pick one problem: for example returns, delays or stock planning.
4. Try an available tool on that problem: measure before and after.
5. Keep control of your data: export, access and clear rules on who uses what.

“We can start using what is readily available to us” was a message from the panel that applies to the warehouse and the webshop alike.

How we work

We are not a warehouse that uses software, but a technology company that has a warehouse. We build our own WMS and digital tools in-house, and our warehouse is a living lab: we test everything on real orders before we rely on it.

For clients, that means real-time stock visibility, integrations with webshops and ERPs, and data that is ready for the next step, whether that is a better report or AI.

Looking for an outsourced warehouse that gives you data, not just space? Request a proposal or explore our 3PL services.