With the kind of year we had so far, who would dare to dream about innovation? A lot of you, it turns out, as is evidenced by the latest installment of our DEL20 innovation cycle. In an edition unlike any other, leading companies from Flanders and Wallonia proved that even a global pandemic can’t stop the pace of progress. Here’s what exciting insights our knowledge-sharing community came up with this year.
But first, let’s briefly recap what DEL20 is all about. “The main idea of the DEL20 ecosystem is to allow businesses to experiment with new technologies like AI, robotics, IoT and more, without having to worry about competition, budgets, or risk of failure,” explains Thierry Bruyneel, partner at delaware and event creator. “Clients can propose a project and we offer them a framework – along with free consulting days for the winner – to help realize their ambitions. Meanwhile, the goal is to share insights and findings with the entire DEL20 community.”
In 2020, for obvious reasons, most of that insight-sharing was done virtually. However, the resulting proofs of concept below show no compromise by any means.
If there’s one thing that keeps mattress fabric manufacturer BekaertDeslee awake at night, it’s production waste. Using Azure Machine Learning, the company developed an algorithm that could help operators predict when a machine was likely to break down or produce a bad batch, based on the vast amounts of historical data that BekaertDeslee had acquired over many years.
Today, BekaertDeslee is reporting up to 20% waste reduction, and the algorithm has been rolled out across 10 plants, good for over 110,000 production orders per year. The UX has been translated into 8 languages and training is underway to get all operators up to speed. You can read more about this project here.
One of Belgium’s leading multinationals, steel wire and coating technologies business Bekaert is always looking for new ways to make processes more efficient. This year, the team explored how rule-based matching and machine learning could make time registration easier, faster, and more accurate.
The idea was to have an Azure-based machine-learning model make suggestions based on the content of the user’s Outlook calendar. As the experiment progressed, it became clear that machine learning was feasible and accurate enough to provide significant benefits. The next step is to investigate the multi-language aspect of machine learning.
How can geotagging solve logistics problems that lead to dissatisfied customers and increased operational costs? That’s what steel producer Joris Ide wanted to know. In their proof of concept, the company decided to experiment with different technologies: RFiD, ultra-wideband (UWB), cameras and GNSS/RTK (GPS).
While none of the above seemed to cover Joris Ide’s business case completely, the company gained a plethora of new insights about these technologies and their possible applications. More than ever, the team is convinced of the benefits of closer cooperation between business and IT, and of exploring technological innovation in a lean and agile way.
One of the greatest challenges sales representatives face is knowing the right time to contact a customer again. Chipboard and MDF manufacturer UNILIN panels decided to explore how a predictive model based on historical client data could help.
Deployed over three phases – data validation, modeling, and testing – the so-called ‘Buy till you die’ algorithm predicts when a customer is likely to make another purchase or whether they will stop being a client, and sends sales representatives a warning. Preliminary A/B testing has already indicated a higher conversion rate. Now, the UNILIN team wants to take things even further by exploring the best channels through which a customer should be contacted.
To attract new and younger customers, multimedia group Roularta decided to explore the possibilities of micropayments via blockchain. Up until now, the group has offered mostly long-term subscriptions to its digital and print magazines, purchased through traditional channels like direct debit, bank transfers, credit cards, etc. With blockchain, they could offer the clients the possibility of purchasing short-term subscriptions while reducing or eliminating the middleman cost.
For Roularta, the most important prerequisites were the stability of the blockchain technology and cryptocurrency, as well as the intuitiveness of the user interface. An initial experiment with Dai, a crypto-coin linked to USD, failed because the transaction costs and the platform’s complexity were too high for customers. Subsequently, Roularta decided to focus on research instead. Today, the group is investigating the possibilities of low value payment providers.
Needless to say, healthcare workers have been under tremendous stress lately. In the context of DEL20, care company Zorgbedrijf Antwerpen decided to explore how smart speakers could help relieve the burden. Their goal was two-fold: to provide relief for overworked healthcare workers while simultaneously offering better care to residents in their facilities.
While the technology already exists, Zorgbedrijf Antwerpen had to think creatively and pragmatically when introducing it in real-life care environments. The main question: which technology is best-suited for residents? With the help of the DEL20 community, the team managed to evolve the idea into a viable concept. Today, Zorgbedrijf is still exploring challenges like language barrier, scalability, and the overall impact on residents and workers.