With more than 100 years of history behind it, Netherlands-based Geostick has grown from a family label printer into one of Europe’s most advanced providers of self-adhesive labels and packaging. Innovation has always been central to that growth, and today the company is applying new tools — particularly artificial intelligence — to take data, efficiency and sustainability to the next level. "Data is very important because, although I think in life you can get very, very far with your gut feeling, it's always good to know, OK, are these changes actually providing the benefit that we thought?" says Tom Schouten, production manager at Geostick. At the Eco-Logic Summit in Milan this November, he and Geostick data scientist Anton Geraedts will share how the company is using predictive AI to fine-tune job planning and reduce waste, as well as generative AI to automate high-volume invoice processing. Both projects, described below, highlight how smarter use of data can improve operations, profitability and environmental impact. Predictive AI for Smarter Forecasting
Geostick is applying predictive models to the mountains of historical job data it has collected over decades. The company has records stretching back more than a decade, sometimes even from presses that are no longer in use, and every order leaves behind valuable information about run times, waste levels and machine performance.
"Basically, we try to use the historical data from all the orders at Geostick to learn how to predict the amount of waste and the amount of time necessary for an order," says Anton (pictured below).
The process begins by comparing those real-world results with the forecasts produced in the company’s CERM system. If CERM predicts that a certain order will take 20 minutes and require 100 meters of substrate, but in reality it took 30 minutes and 130 meters, the gap between forecast and actual is logged. Multiply that by hundreds of jobs every day, and the inaccuracies become significant.
Geostick also uses Power BI, Microsoft’s business intelligence and visualization too, to turn raw machine and production data into clear dashboards that highlight where adjustments are needed most.
What makes predictive AI powerful is its ability to spot these patterns across thousands of orders and make systematic adjustments to the parameters inside CERM. Instead of relying on gut feeling or occasional manual tweaks, the AI can recommend changes that bring forecasts closer to reality. For example, one finishing process had long been assigned an extra 300 meters of waste material as a default setting, but the historical data showed that this consistently overstated the actual amount required. By adjusting the parameter downward, Geostick could more accurately match expectations to results — and save significant material in the process.
These improvements ripple outward into the rest of the operation. With better forecasting, job schedules can be planned more confidently, press operators know what to expect, and managers can sequence jobs in ways that minimize downtime or unnecessary cleanups. The goal is to create what Tom and Anton describes as true predictability.
Generative AI for Invoice Automation
On another front, Geostick is using generative AI to tackle one of the most repetitive and resource-intensive tasks in its business: processing more than 100,000 invoices each year. Traditionally, that work has required about 10 full-time employees to manually copy details from PDFs into the company’s CERM system — identifying the client, matching the order, and entering the quantities, dates and other critical data point by point.
Now, an AI-powered application built in-house takes over much of that process. Geostick’s system takes a client invoice in PDF form, runs it through ChatGPT and extracts the key details — such as which client and order it belongs to, what was purchased, the quantities and the delivery date. In other words, all the steps a human would normally enter manually are now handled automatically by the AI. Once the AI has extracted and verified the information, it automatically creates a new order in CERM. Human staff still double-check the results, but the time savings are already clear.
For Geostick, automating invoice processing frees skilled employees from repetitive tasks, allowing them to focus on higher-value projects in production, planning and customer support, Tom says. In an industry facing ongoing challenges around talent and staffing, the ability to redeploy human energy where it’s most impactful is as valuable as the time saved.
Data, Sustainability and the Future
Although efficiency is the immediate driver of these AI initiatives, sustainability is an equally powerful outcome. Every meter of material saved through more accurate forecasts represents less waste heading to recycling or disposal. Every hour of unnecessary press time eliminated means less energy consumed. And every invoice processed automatically rather than manually reduces the risk of errors and rework.
For Anton, sustainability is embedded in almost every data-driven improvement. He described one project where the company used order frequency and shipping analysis to optimize transport schedules. By consolidating shipments and matching them more closely to customer needs, Geostick reduced CO₂ emissions while also making life easier for employees.
That alignment between operational gains and environmental benefits is central to Geostick’s long-term vision, Tom says, and it’s exactly why the company will be a featured participant at the Eco-Logic Summit in Milan. The event will bring together print leaders from across the region to explore how sustainability and AI can work hand-in-hand to transform the industry. For Tom (pictured below), the message to his peers is simple: “We’re trying to show how AI can work for print in very practical ways. If we make operations more efficient, the byproduct is sustainability.” For the next generation of print leaders, that dual focus — better business and better environmental outcomes — is the way forward, he said.
