Interview series
02
Date: 2024.11.12
“Designing an experience together” rather than simply “creating AI.”
Roland
Kazuyoshi Sasamori
Roland Corporation: Leading Manufacturers and Distributors of Electronic Instruments in Japan. Insights from Kazuyoshi Sasamori on our collaborative project, type: entry-hyperlink id: 3DsFEexe7RUer5hTi6S9Yy to measure match between presets and phrases.
Introduction
Can you introduce yourself and explain your role in the company?
My name is Kazuyoshi Sasamori. I am currently part of the Roland Future Design Lab Advanced R&D Group which to utilize new technologies such as AI and Web3. As of May 2024, I became AI Product Manager, where I lead efforts to incorporate these technologies into our products and services.
Are there other AI Product Managers like you?
Actually, there aren’t any yet. We’ve entered the phase of practical implementation, so for now, they’ve decided to appoint just one person. There was feedback from the overseas team that it was hard to grasp the responsibilities of this role if it was framed as an expert position. So, for AI-related matters, we’ve consolidated everything under this role for the time being.
Can you tell us how you came to know about Qosmo?
In our technical research division, we conduct activities to understand trends in AI and digital technologies, where I was primarily responsible for AI. At that time, Tokui-san’s AI DJ project was already well-known, so when it came to Japanese AI companies, especially those involved in music, that’s when I first learned about Qosmo.
About projects
Can you tell us about the issues and challenges you were aware of when starting the project?
At the beginning, we vaguely thought there would be a correlation between tones and phrases, and we wanted to take advantage of that. While we had a general sense that AI could be applied in some way, the biggest challenge was that, to be honest, we were completely in the dark about which methods to use and how to build the system.
After that, you reached out to us, and we had a meeting, right?
Yes, that's right. Instead of the currently trending “text to something” approach, I had a feeling that a more multimodal approach would be necessary. So, I started thinking, “Which company would be good at that?” The first company that came to mind was actually Qosmo. Since there were many aspects that we couldn’t figure out on our own, you initially proposed the method, and from there, you kindly explained how we should prepare and the steps we needed to follow. That made it much easier for us to get an understanding of the overall process from preparation to execution.
Can you share any results or feedback you received?
One of our goals was to build a model as part of a verification process before starting prototype development, and we were grateful to reach that goal in such a short period of time. Once we had that, we were able to show it as something tangible to the people around us, which made it easier for them to understand compared to just explaining it with words. I think the best part of working together was being able to deliver such outputs.
What were the difficult and enjoyable aspects of advancing the project?
The challenging part was that the project had a relatively tight schedule, so I felt it would have been better if we had more time for discussions. That may have been the most difficult aspect. However, we had various communication tools, so we were able to resolve a lot of issues through conversations.
As for the enjoyable part, it was not just about asking for the technology to be implemented, but rather working together to discuss how we could achieve our goals and what results we could expect. It was great to have a collaborative approach, actively engaging in the process together. That’s something you don’t often get to experience, and I found it very rewarding.
Lastly, can you share your thoughts on reflecting back on the collaboration/project with Qosmo?
I think part of it may be because it's related to music, but I appreciated how we didn’t stick rigidly to a specific technology. Instead, we exchanged ideas to make something more interesting, and we could provide feedback to each other along the way. I really enjoyed that aspect. My biggest hope now is that we can continue this kind of collaboration, making things even more interesting and ensuring that what we create reaches the end users.
We completely agree, and it’s delighted to hear that!
What left the deepest impression on me was that it felt more like we were “designing an experience together” rather than simply “creating AI.” Having those discussions from that perspective gave me a real sense that we were both equally involved in the development and research process, which I found very rewarding. I really appreciate it, thank you.