Since September 2017, the Harmonic Engineering Laboratory of Hokkaido University and TMJ, Inc. of the SECOM Group have been jointly developing an algorithm for automating shift adjustment.In the future, we will verify the algorithm for automation of shift adjustment work linked with the chat system that utilizes AI operated by TMJ.The effect has been demonstrated in a trial, and we are aiming for practical use by the end of FY9.
Currently, in labor-intensive industries such as the service industry and the restaurant industry, the shortage of human resources is becoming more serious, and shift adjustment work in the event of sudden absenteeism is a heavy burden.Therefore, we jointly developed an algorithm to reduce the burden of shift adjustment work with the Harmonic Engineering Laboratory of Hokkaido University, which is researching information science.
The algorithm developed this time extracts a list of target people as to how many people should be called out for sudden vacancies, determines when and when to call out, and optimizes it.In shift adjustment so far, it depends on the "experience and intuition" of each person in charge, and since the adjustment know-how is not accumulated, it has been personalized, but this algorithm reduces the man-hours required for selecting alternatives and productivity. And improve the acceptance rate.
In addition, the AI that learned the algorithm and the chat system are linked, and a message is automatically sent to confirm whether or not attendance adjustment is possible.Final adjustments will be made by manned chat to the staff who responded.By using LINE from the telephone, the shift adjustment channel streamlines contact with the target person, reduces the work burden, and eliminates the psychological burden of making and receiving calls.
Looking at the effect simulation based on the results of this trial verification, in the case of a target list of 100 people, the number of contacts is 50 for telephones, and 75 for chats, increasing the chances of negotiations by improving contact efficiency.The number of people who accept work is 7.5 people by phone, and 9.1 people by chat, which improves the acceptance rate by selecting the most suitable target person.Furthermore, the man-hours (hours) per response were 42.3 minutes for chat, compared to 10.3 minutes for telephone, which reduced man-hours and improved productivity.
Currently, in collaboration with Hokkaido University, we are aiming to improve the accuracy of the algorithm while continuing learning based on the results of chat answers, continue to verify and respond to issues, and put it into practical use by the end of FY2018.
