CASE Achievements

  1. HOME
  2. Achievements
  3. Construction of automatic ordering system for major retailers

Construction of automatic ordering system for major retailers

Construction of automatic ordering system for major retailers

Build an automatic ordering system using general-purpose AI. We undertake operation and maintenance including AI learning support after operation.

Participated in the construction of an AI automatic ordering system for hundreds of stores

The supermarket of major retailer X has introduced an advanced automatic ordering system using AI technology. AI forecasts demand for each store and product by adding information such as weather forecasts, local event schedules, and search trends to the number of products in stock and sales. The number of orders will be calculated accordingly (Fig.). This is an initiative to reduce the burden on the person in charge of ordering and improve the productivity of the site.

CUBE SYSTEM started construction of this system from the latter half of 2017. It was appointed in recognition of the track record of being involved in the maintenance of the core system of Company X for many years.

In 2018, trial operation of the system will start at some stores. The number of stores that have introduced it has been gradually expanded, and it is now officially operated at all stores. Of these, the operation and maintenance work at 100 stores is also the role of CUBE SYSTEM

Data linkage unit that draws out the capabilities of low-cost general-purpose AI

The AI engine that forecasts demand at the core of the system was not developed exclusively. By using a general-purpose AI solutions Services we were able to achieve speedy and low-cost development without the need for construction support by experts such as data analysts.

For the same reason, AI is not designed to read company X's business data directly. Moreover, the core system of Company X, which has a long history of operation, is the so-called "legacy system". The data obtained from it was in a cumbersome format.

The data linkage department solves this problem. Product data received from the core system and weather prediction data received from a third party are converted into a feature set (numerical group for learning). Pour into AI.

Being a pioneer in developing with ETL tools

The ETL (extract / transform / load) tool used for data processing and editing was used to develop the data linkage unit and its interface.

Development with ETL tools with an intuitive UI can be done efficiently once you get used to it. It has been attracting attention recently because it can dramatically speed up development compared to coding using general development languages such as Java. However, when the CUBE SYSTEM first came into contact in 2017, the documentation was still undeveloped, user there was almost no Japanese information. Of course, this is the first ETL tool that customers Under such circumstances, while sharing information among the participating members, I tried to master how to use it.

In the current system development, not only the use of AI but also the need to process a large amount of data is increasing. The low-code ETL tool development know-how that makes this possible efficiently can be expected to be applied to various projects in the future.

Contracting maintenance process that is close to the customers

A general-purpose AI engine cannot make highly accurate demand forecasts using only the data read during system construction. In fact, since the trial operation started in 2018, there have been a series of complaints from stores that have introduced it for a while. However, AI is learning every day and refining its predictive model.

The demand forecast calculated by AI is first sent to the function that automatically calculates the number of orders. Therefore, processing is performed according to the characteristics of each product, and the results are sent to each store. At the store, the number of orders received is modified by the person in charge based on experience, and the final number of orders is decided.

While responding to inquiries that occur during the maintenance process and hearings with stores that have introduced this system, we will incorporate what can be systematized each time. customers maintaining the system in step with the customer's retail site on a daily basis, the system itself will evolve along with the prediction accuracy.

We are fully responsible for the maintenance process that is close to customers

Voice of site staff

This is a project that we have been working on while learning new technologies together with our customers In the development of new functions in the operation phase, we discuss with the end user customers and define the requirements. We almost take care of the process of modifying and implementing the functions. We consider it a proof of trust.

At our company, the role of our field members is user work in a well-balanced manner. This project has made it possible. We are gaining important knowledge every day, such as development with ETL tools and AI learning support. Combined with our multifaceted efforts, we should be able to develop in various ways. We will continue to catch up with new technologies so that we can further increase our presence in the advanced technology field.

  • 1st solutions Business Headquarters
    KM
  • 1st solutions Business Headquarters
    YA

Please feel free to contact us.

Inquiry