Time was spent responding to routine inquiries
Our company has set up an in-house call center to respond to inquiries about in-house facilities, PCs, mobile phones, etc. All inquiries were handled by operators, but many of them were routine inquiries and a lot of time was spent responding to them. As a result, there was a chronic shortage of manpower, and the issue was a declining call answering rate. We thought that by automating routine responses, operators would be able to focus on the inquiries they needed to respond to, so we considered introducing a voice bot.
High recognition rate and scalability through external linkages also highly rated
While considering a voicebot, we learned about "AmiVoice ISR Studio" during a webinar. During a trial run, we found that the recognition accuracy was good, and we were able to easily create scenarios in-house to suit the usage scenario. We also evaluated the fact that it has many functions to further improve recognition accuracy, such as engines prepared for each purpose and a word registration function, and decided to introduce it. Another reason for introducing it was that it has an external integration function, which we expected would enable flexible telephone response by linking with external systems.
Automate over 20% of inquiries and provide 24/365 support
When we used "AmiVoice ISR Studio" at our in-house call center, we were able to automate routine responses, which account for more than 20% of inquiries. This allowed us to focus on inquiries that required an operator's assistance, improving work efficiency and call answering rates. In addition, by utilizing the SMS sending and email notification functions, we are able to provide information to users quickly. We feel that another benefit is that inquiries that can be handled automatically can be handled 24 hours a day, 365 days a year.
Users have commented that the process has been reduced because they can now make inquiries by phone without having to search through FAQs.
Integrating voicebots and generative AI
Starting in October 2024, we will combine generative AI and RAG (a technology that improves response accuracy by combining generative AI with external information searches) to further increase the proportion of automation. Currently, only some inquiries, such as inquiries when receiving suspicious emails and inquiries regarding error handling, can be automated. By linking generative AI with RAG, we will be able to respond to frequently asked questions and we believe that we can automate approximately 10% of inquiries.
Towards realizing automated responses that resemble natural conversations with people
Current voice bots follow fixed scenarios like IVRs, and are not like natural conversations with people. By combining generative AI with voice bots, we believe that we can create automatic responses that are as natural as conversations with people.
Mitsubishi Electric Information Systems Corporation (MDIS) is a system integrator (SIer) that provides solutions to resolve corporate IT issues.
Through this initiative, we have been able to automate over 20% of inquiries, improving response quality and streamlining operations. Based on these results, we will propose similar solutions to our customers and contribute to solving their problems.
Data Strategy Group Takashi Yamagata
