Artificial intelligence works best when tailored to a specific business case. Demand and inventory planners in FMCG therefore need specialized AI to support them in making decisions about key processes, says Kamil Folkert, CEO of Occubee S.A., co-creator of the SaaS platform for dynamic inventory management based on AI.
PORTAL SPOŻYWCZY: EVERYONE IS TALKING ABOUT THE USE OF ARTIFICIAL INTELLIGENCE IN THE SUPPLY CHAIN TODAY, BUT CAN EVERY FMCG COMPANY USE IT AND MAKE THE MOST OF IT?
Kamil Folkert: In the Polish FMCG market, both among retailers and manufacturers, it is easy to identify companies that are champions of change in this area. Most often, these are large entities that recognized years ago that data is money, and advanced data analysis based on AI is a catalyst for optimizing processes and increasing business profitability. Other entrepreneurs not only want to follow the path paved by leaders and adapt these solutions for themselves, but they also know that this is the only way to stay competitive in the market. I am convinced that each of them can benefit from the potential of AI, albeit on a different scale and in various operational areas.
WHERE SHOULD THEY START? IN WHICH PART OF THE BUSINESS CAN ARTIFICIAL INTELLIGENCE BRING THE MOST BENEFITS?
In my opinion, the priority process in the FMCG industry that is worth optimizing with AI is demand forecasting. This is primarily because high-quality forecasts are a prerequisite for optimal inventory management in stores and warehouses. Securing an inventory level that matches consumer demand during a given period is the Holy Grail for retailers and manufacturers. It has a significant impact on cash flow, sales performance, service levels, and many other business parameters. Errors made at the forecasting stage create a snowball effect, accumulating and negatively affecting processes such as production planning or store replenishment. Artificial intelligence helps avoid these errors.
HOW DOES IT WORK?
AI-driven forecasting uses advanced algorithms that analyze historical data and take into account additional variables, such as sales seasonality, consumer trends, weather forecasts, holidays, and special events. Machine learning models can process vast amounts of data and information across the entire product range, sometimes involving thousands of SKUs, which is simply impossible for humans to do. Additionally, as the name suggests, they are capable of learning, allowing them – continuously powered by new data – to account for emerging trends and generate forecasts that are well-suited to changing business conditions.
DO SUCH MODELS WORK IN OCCUBEE?
Yes. The core of Occubee, our AI-based dynamic inventory management platform, consists of proprietary forecasting models that are specifically tailored to the FMCG sector and particular business cases. We parameterize and train them using our clients’ data, in a very detailed manner, i.e., individually for each product-location combination. This provides the best results for the forecasts we generate, and subsequently, for store or warehouse demand, picking recommendations, and supplier orders. Each of these steps can be performed automatically by the system, which, in practice, means that we enable the automation of a process that requires making millions of decisions every day.
THIS SOUNDS COMPLETELY DIFFERENT FROM THE ARTIFICIAL INTELLIGENCE WE KNOW AS CHAT GPT.
Definitely. Artificial intelligence is a very broad term, but our approach fits into the concept of so-called narrow AI, which is specialized and focused on very specific, concrete tasks. This is not generative AI, powered by content copied from the internet. This distinction is also highlighted by Gartner experts, who explicitly point out that GenAI is not a cure-all and that in areas like forecasting or planning, it has limited use. I am convinced that narrow AI is the one that can bring the greatest business benefits to retailers and FMCG producers, especially since we are talking about optimizing their core processes.
SINCE WE’RE TALKING ABOUT BENEFITS, WHAT IMPACT DOES AI-BASED FORECASTING HAVE ON BUSINESS?
Let me start with an example of a manufacturer. Medium- and long-term demand forecasts, aggregated weekly or monthly, enable better planning and production management. They allow for proactive inventory management and adjusting the production cycle to market demand in advance. As a result, fast-moving consumer goods manufacturers can reduce overstocking, thus freeing up frozen capital. On the other hand, they minimize the risk of out-of-stock situations, which lead to a lower service level. High-quality forecasts help experts make business decisions that result in more efficient resource use and cost reduction.
AND WHAT ABOUT FMCG RETAILERS? HOW DOES IT WORK FOR THEM?
The benefits are similar, though in this case, we primarily rely on short-term sales forecasts, generated daily, for each SKU-store combination. With the resulting store demand calculations, followed by picking recommendations, retailers can ensure adequate product availability in stores. This, in turn, reduces the risk of lost sales due to out-of-stock situations while maximizing the chances of increasing product turnover and boosting sales. It’s also worth mentioning that demand forecasting helps minimize losses, which is a major issue in retail.
SMALLER WASTE IS NOT ONLY A BENEFIT FOR THE BUSINESS BUT ALSO FOR THE ENVIRONMENT.
Exactly. For demand planners in the FMCG sector, fresh products are a challenge. On one hand, they have a short shelf life, and on the other, they often require special storage conditions. They are prone to spoilage, which means a high risk of waste. According to the 2024 Food Waste Index Report by the UN, in 2022 alone, global food losses in the retail sector amounted to a staggering 131 million tons. This is equivalent to the weight of over 21 Great Pyramids of Giza or 359 Empire State Buildings. Inventory optimization based on demand forecasting is one of the ways retailers can reduce this waste and, in turn, meet environmental goals.
The interview was published on Portalspozywczy.pl.
