How quickly can you put together a computer using the configurator?
Appetite grows as we eat. When we traveled to Kanardi for our first meeting a year ago, neither we nor our client anticipated how much our cooperation would develop. It's hardly surprising, because as the first tests showed, the time for a vendor to prepare a statement of a new computer can be reduced by up to more than 96%.
Beginnings of cooperation
As a professional supplier of IT equipment for business, Kanardi, along with its growth, accumulated an ever-growing database of components that made up their solutions. Each custom-built computer was dozens of key components plus hundreds of minor ones. Each could come from a different manufacturer and distributor, each had slightly different specifications and different prices. At some point, the company's management came to the conclusion that working with traditional spreadsheets was already taking up too much time and that this time would be better spent on new projects.
And this is where we came in with our product configurator - CONFIGURAT.IO, which turned out to be the perfect answer to the company's pains. Our implementation was even partially financed by the "Innovation Voucher" - a special grant from the European Regional Development Fund, which is intended for the development and implementation of innovative solutions in micro, small and medium-sized enterprises.
When we developed the configurator and put it through its first tests, we found that with this solution, the time to configure a custom-cut product was reduced from 120 to 4 minutes!
New challenges and further plans
Currently the system is ready, but it needs to be fed with data. This part of the implementation was eventually handed over to us as well. It turns out that uploading and, most importantly, unifying intermediates from multiple suppliers is as challenging as building a system to manage them at the design stage of the final product.
Seven suppliers with thousands of products on offer. Each has its own naming system, its own methods of description, other file formats. On top of that - you know - computer technologies are constantly changing, and there is a steady stream of new releases to the market. After a preliminary calculation - about 200,000 indexes to be processed, unified and bundled. That's more or less what our next project for Kanardi looked like.
Biggest challenges
Challenge one - the same product/item (e.g., a given processor) can be supplied by multiple distributors under the same name. The system must recognize this and name the product in such a way that it is uniquely attributed to the distributor - recognize the source of the goods and it is from there that the product is imported into the configuration along with the price and description that belong to the given supplier.
Challenge two - the opposite of the first - there is a lot of overlap between suppliers in product/element names, parameter names, description styles, etc. Two identical products are described in such a way that the system might not recognize them as identical. Therefore, care must be taken to catch the differences and find one common way to describe them. For example - the length of an identical cable could be specified as 0.5 m or 0.5 m or 50 cm. For a human it is the same, but for an algorithm - no longer.
Challenge three - suppliers update their data in different ways. Some do it periodically, others occasionally. Some use an xml gateway to do so, others provide integration through a dedicated API, others through csv/xls files (streamed or FTP downloaded).
So, you need to configure the system so that it takes into account multiple standards of behavior. What's more - to interpret the list of indexes or categories in the same flawless way every time, and to correctly pull data into the configurator. Although there are many vendors, and each has its own way of presenting data, the whole thing is to create a homogeneous and - above all - error-free set with clear and unambiguous descriptions.
But there are also minor challenges - some data needs to be glued together, and others better separated. Some sub-suppliers' systems are fully integrated with their warehouse system, others are not. This is compounded by today's frequent price changes and lengthening delivery times for some components.
"It's hard to imagine such a complex project without the involvement of artificial intelligence," says Michal Klin of EXSO, who works on solutions for Kanardi. - The large amount of data and the need to recognize it requires learning and continuous improvement, and this challenge is just right for AI-based algorithms. It can be said that all three of us are learning together on an ongoing basis - the customer, us and.... our AI!"
It looks like there will be even more of this learning, as work is already underway on more projects at Kanardi.
