A catalogue that had outgrown paper
JORI's collection runs to 102 base models, 151 types of leather and 160 fabrics, plus finishing options for feet and armrests — and customers can mix components to fit a particular room. The combinations run into the thousands.
That is unmanageable on paper, and it was becoming unmanageable in a showroom too. A salesperson could not hold the options in their head, and a customer could not picture the result.
We're seeing a new way of browsing, especially for high-end products. Before, we offered customers a catalog; it was enough to describe the product using words and images. Now, consumers want to see our furniture in 3D. It's like buying a car — they want to be able to walk around the product and view it from all angles.
Wim VantommeExecutive Commercial Manager, JORIThe configurator had to know the manufacturing rules
A standalone 3D tool would have been the easy build and the wrong one. JORI's manufacturing equipment already works from detailed 3D models made by its own designers, and the business rules that decide which component combinations are possible — and which would simply be uncomfortable — had lived on IBM i since the 1980s.
So we put the configurator on the same machine as the rules. The 3D object database sits in Db2 for i. RPGLE web services talk JSON to the front end, which is built in Unity, the games engine, compiled for IBM i using Chroot. Ghostscript and ImageMagick produce the print-outs, Substance renders convincing fabric, and Drupal runs the site around it. The whole stack runs in one partition on JORI's existing Power System S814.
What's really cool is that this innovative, open-source solution is fully integrated with application code and business rules that our client has entrusted to i since the 1980s. Past investments are protected and there are no restrictions on what can be achieved with new technology.
Koen DecorteCD-InvestBecause the configurator asks the manufacturing system rather than a copy of it, it cannot produce an order that the factory has to query or reject. That single property is what made everything downstream possible.
What it changed
The configurator won the 2016 COMMON Benelux Excellence Award in the application development category. It is embedded in the websites of more than 230 furniture dealers across Europe and runs as a mobile app used by over 3,000 people — consumers and in-store advisers with tablets.
- 95 % of retail partners now use it
- 50 % less time from customer order to final delivery
- 80 % less time to market — six months down to a few weeks
- Some items now produced in five days that previously took eight weeks
Because a configured order flows straight into manufacturing, JORI can also list models it has never physically prototyped. They comply with the business rules, so they can be built; if customers configure and order them, the factory makes them. No prototype, no reprinted catalogue.
Thanks to the configurator on IBM i, dealers can offer our full range of furniture in a quick, easy, and accessible way. In addition, they can instantly give customers accurate pricing no matter how complex the chosen options.
Wim VantommeExecutive Commercial Manager, JORIThen March 2020
JORI and CD-Invest had begun thinking about a new production planning system in February. The existing one was a simple calculation algorithm inside the ERP, and it could not express what was about to be required. When the lockdown came in March the project was accelerated rather than paused, and it was live by the end of April.
The new plans had to hold constraints no planner had needed before: social distancing expressed as positions on the factory floor, a 72-hour delay when a piece passes from one team to the next, and leather — which cannot be disinfected — held in controlled conditions before handling.
It runs on the same IBM i and Power Systems platform as the configurator, drawing on configurator data alongside the ERP. IBM Services helped tune the SQL. The factory reopened quickly and legally, and JORI's owner was confident enough in the result to invite the Belgian government in to inspect the process.
Where it went next
- An outlet store using JSON and Db2 for i, integrated with IBM Watson Language Translator so dealers type in their own language and the text appears in German, French, English and Dutch
- IBM Watson Assistant suggesting products by region — smaller pieces where homes are smaller
- Predictive models in scikit-learn, Chainer and H2O Driverless AI to choose where to put the next physical store
- Architects and interior designers importing the 3D models into their own visualisation software
Sources: cdinvest.eu portfolio, JORI · “JORI Looks to the Future by Investing in Data Analysis with CD-Invest”, 13 December 2020 · “CD-Invest helps JORI increase manufacturing efficiency while addressing COVID-19 concerns”, 27 July 2020 · IT Jungle, 17 March 2021. All quotations as published.