You can also find us on the official ESTsoft LinkedIn.
ESTsoft, an AI service specialist company, is creating a better daily life through AI under the slogan of ‘MAKE YOUR LIFE EASIER WITH AI’. Unlike chatbots that merely exist on a screen, Perso Interactive, the core business, creates Physical AI that helps with necessary tasks such as guidance, recommendation, and interpretation by facing people in real spaces like kiosks, vehicles, and stores.
Perso Interactive, which prioritizes "interaction" with people, is being actively utilized in various fields such as exhibitions, retail, tourism, and construction. With its dazzling technology and naturalness, its range of application and potential are growing remarkably, placing it at the center of ESTsoft's AI business.
AI creates the moment when it faces people.
In this edition of ITSELF, we met with the Perso Interactive Business Division, which is spending busy days with barely any time to breathe. The Perso Interactive Business Division is further segmented inside according to roles. To hear a richer story, we met with three people: Kim Hyun-jin from the Advanced Technology Team in the Human Department of the Planning Center, Kwon Dong-young from the Japan Business TF and Solution Development Team of the Development Center, and Hwang Su-young from the AI Content Business Team of the Business Center. From discovering customer problems and designing conversations that people will naturally accept, to implementing those experiences on actual screens and devices, we listened to the stories of these three people, each with their own unique roles.
Q. Hello! First, please briefly introduce yourselves and what kind of work you do.
Hyun-jin) Hello. I am Kim Hyun-jin from the Advanced Technology Team in the Human Department of the Perso Interactive Planning Center. I turn the AI avatar roles desired by customers into actual conversations. I design and verify things like what to answer to which questions, what tone of voice is appropriate, and whether the same impression can be given across multiple languages.
For example, in environments with many variables like exhibition halls, because not only prepared questions come in, the process of creating tests that resemble real situations and repeatedly having conversations is very important. I do the work of finely polishing every detail down to the last sentence that the user hears.
Dong-young) I am Kwon Dong-young from the Solution Development Team of the Perso Interactive Development Center and the Japan Business TF of the Business Center. Based on on-device AI technology, I create products that allow AI avatars to react to human speech in real-time and perform necessary tasks.
I have been involved since the period when the potential of Perso Interactive was first being verified, and I built the foundation for the SDK through which the AI avatar reacts in real-time within the device. Since then, I have expanded the boundaries of the product by commercializing internal technologies into public SDKs and connecting AI conversations to actual actions in customer service.
Recently, in a Japanese mobility project, I have been creating a service where foreign passengers and Japanese drivers can converse naturally, and receive guidance on nearby attractions and local information in their respective languages while moving. Making an AI agent that helps with the entire process of conversation and movement so that people can communicate comfortably and get necessary information without language barriers even in an unfamiliar city is my job.
Su-young) Nice to meet you! I am Hwang Su-young from the AI Content Business Team of the Business Center, meeting customers at the very front line. There are really many problems that can be solved with AI avatars, but it is not easy for customers to think of how to utilize them first. So, I ask customers, "What difficulties do you currently have in your work field?" and discover the problems they are experiencing. Then, I draw up a solution to solve that problem with Perso Interactive. Changing that answer into a problem that the Planning Center and Development Center can solve together, and ultimately connecting that idea to actual projects and launches, is my role.
Q. There might be people who still cannot visualize exactly what Physical AI and AI avatars are. Could you explain in more detail what Perso Interactive creates?
Su-young) In a word, it is creating a new way of responding in spaces where people are present. You can meet them in various forms, from stations (kiosks) that welcome visitors at exhibition halls and stores, to AI avatars that consult, educate, and converse on PC webs, and services that provide guidance or interpretation through everyday devices like tablets.
Perso Interactive is changing many parts of daily life. Representatively, at exhibition halls or general stores, visitors can ask the AI avatar questions they are curious about without burden, even without calling an employee, and employees can reduce repetitive guidance resources to focus on more important things. Even two people speaking different languages can continue their conversation without discomfort if they have Perso Interactive next to them.
Hyun-jin) However, what we make is not limited simply to superficial parts like information delivery or conversation. This is about creating people's 'experiences'. Even with the same information, the design completely changes depending on who says it, where, and to whom. For example, in a flashy exhibition hall, you must speak shortly and clearly, and in a store, it must be friendly but not burdensome. If it is an avatar representing a brand, not only its appearance but also its tone of voice and reactions must be characteristic of that brand.
We do not stop at whether the AI says the right answer, but always think whether people can accept that answer naturally.
Dong-young) And we connect all these designs so they are implemented seamlessly on the screen being faced. If you say, "Show me the nearest store," a map screen opens, and if you say, "Recommend a light product suitable for sensitive skin," it proposes appropriate products based on the concerns and conditions contained in natural language. Among them, if you pick a product you are curious about, it connects directly through images and videos so you can see its details.
In other words, the AI avatar created by Perso Interactive does not just converse well, but leads users to the final actions of exploring, comparing, and selecting right on the spot.
A Perso AI Station is installed on the first floor of the ESTsoft Seocho office building.
The Japan Business TF connecting different languages into one
Dong-young added more about Perso Interactive's Japanese business, which the Japan Business TF is working on. While the number of foreign tourists visiting Japan is growing rapidly, local labor shortages and aging populations are intensifying. Rapidly capturing these changes and seeing the possibility of a new market, the Perso Interactive Japan Business TF, starting with taxis, is creating a service that combines AI avatars with real-time interpretation so that local Japanese people and foreign tourists can converse naturally anywhere during travel.
Dong-young) In an actual taxi, it is difficult for a driver to look at a screen for a long time while driving or operate a device every time for interpretation. It can be dangerous. So, we created a new experience of interpretation where the conversation continues without any special button manipulation, and the passenger screen and driver screen show the same content in their respective languages. We finely tuned it so that the conversation does not cut off and continues smoothly even amidst vehicle noise or unstable communication speeds.
The most important thing we thought about while making this service was 'conversation between people' rather than the technology itself. We wanted to create an experience where drivers could communicate comfortably with passengers while concentrating on driving, and naturally exchange necessary information during travel. First, running this taxi service well is our first goal, and next, based on what we obtain from this service, we want to expand to general stores, hotels, tourist guidance, etc., finding places where Perso Interactive can help so that anyone can use it naturally.
Q. If I travel to Japan, I would definitely like to meet it. Other than this, are there services where we can directly experience Perso Interactive's technology?
Dong-young) Yes. We released Perso LiveDub and Perso Live Translator so that more people can directly experience the real-time conversation technology developed by collaborating with many companies. LiveDub is a service that allows you to converse in another language while preserving your voice and lip movements using dubbing technology, and Live Translator plays the role of translating 10 languages in real-time to connect the conversation of two people facing each other. It can be usefully used when conducting global video conferences or having conversations with foreigners. Technologies accumulated through global exhibition halls, overseas stores, and the Japanese taxi business have extended to work and daily life areas such as video conferencing.
Q. Then, what is Perso Interactive's unique strength?
Hyun-jin) It is the ability to consistently create the impression of a person or brand from the avatar's appearance to its voice, tone, and conversation. We directly film the actual model of the avatar in our in-house studio, and constantly refine the quality until it reacts naturally to user requests. We can design from start to finish as the customer wants, such as what the avatar looks like when speaking and what kind of impression it wants to leave.
Dong-young) The same AI avatar can be deployed in various environments, from personal devices like iPhone and iPad to Mac mini and large stations. Since the self-developed STF engine creates lip shapes and facial expressions in real-time to match the voice, and performs core processing within the device, quick response and flexible installation are possible.
Su-young) Perso Interactive becomes a professional promoter of our company in exhibition halls, an employee recommending products in stores, and a show host in live commerce. As mentioned earlier, it also becomes a dedicated interpreter and guide in Japanese taxis. Connecting to screens already familiar to customers, such as personal devices, kiosks, and vehicles, and combining languages, voices, knowledge, and functions, the ability to create diverse roles as desired by each brand is the unique competitiveness of Perso Interactive.
Perso Interactive can be deployed across all platforms from personal devices like iPhone and iPad to Mac mini and large stations.
Q. It seems that creating a new market would still not be easy.
Su-young) The most difficult point is that customers do not have a completed answer yet either. Unlike when purchasing ready-made products, it is difficult to just compare spec sheets, and it is not easy to imagine what will change before actually seeing it installed.
So, when talking to customers, we talk with specific scenes or numbers rather than technology names. Like, "Reduce repetitive inquiries at the entrance by 60%", "Customers easily find products in their own language", "One show host produces 3 times more content". Even if it is a small PoC, the process of letting customers see and experience it directly and then reading the reaction together is important.
Hyun-jin) Another difficult point is that human conversation is much more free than expected. Sometimes users do not speak questions to the end, and they ask the same meaning in completely different expressions. There are also moments when surrounding noise is severe or multiple languages are mixed.
So, preparing a few correct answers is not enough. We must collect questions received in actual fields and repeatedly converse in multiple languages to find out at what moments it becomes awkward. Even a small tone of voice, length of answer, or interval of silence can be felt greatly by humans.
Dong-young) In the development field, we must think more about when things don't go as expected than when they go well. The exhibition hall network might be unstable, or the app might stop unexpectedly. It must start by itself without human manipulation, and even if a problem occurs, it must return to its place by itself without a person in charge running to it every time. Still, it's fun. (Laughs)
Q. Since you say it's still fun, on the contrary, I'm curious about what is good about making a product where the correct answer is not set.
Hyun-jin) As Su-young said, because there is no completed answer, it is great that small reactions from users connect directly to learning points. Feedback such as "The appearance before talking is a bit awkward" or "I wish the answer was a bit shorter" changes the avatar's movement, tone of voice, and test standards. The experience where hypotheses set at the desk change into better answers upon meeting people is really fun.
Su-young) Being able to create the language of the market together with customers is also attractive. The conversation function, which started as simple guidance at first, is moving forward into more specific and broader fields such as product exploration, live commerce, interpretation, and consulting. When we properly solve one customer's problem, other industries also see a new path, thinking, "We could use it like this in our field."
Dong-young) I feel the same way. I like that we don't leave the answers discovered here in only one project. Like showing maps or images, common movements of avatars, and remote operations, I think it is meaningful that we can productize what we made in one field so that the next customer can use it from the beginning.
Q. How will the Physical AI market change in the future? What role will Perso Interactive play in it?
Dong-young) When people hear Physical AI, they often think of robots that move like humans first, but I don't think it necessarily means only robots with physical bodies. In exhibitions, stores, hotels, vehicles, and public spaces, if AI understands surrounding situations, reacts to people first, converses in natural language, and performs actual tasks, that is also an important Physical AI. In the future, rather than interfaces where users find set menus and press buttons, interfaces where AI understands situations first, converses in multiple languages, and executes necessary tools will become more common.
Currently, Perso Interactive binds the most familiar avatar to humans, its own on-device technology, connection with customer systems, and field operations into one product, playing the role of deploying that interface into reality. Now, we will expand the experience accumulated domestically into mobility, retail, and hospitality environments in various countries. We will reflect local languages, service methods, and partners' operating environments into the product to create a model that can be used naturally in any country.
Su-young) So in the future, AI will go deeper not as a kind of exhibit, but as one of the ways of working in stores, exhibition halls, hotels, and the mobility industry. Through Perso Interactive, customers will directly feel "So our working environment has changed" rather than the fact that "We introduced AI". I think our role is to become a bridge where success in this field leads to adoption in other spots and industries.
Hyun-jin) In such changes, Perso Interactive will allow AI to approach humans as a more natural and familiar existence. An AI that, rather than simply speaking a lot of information, speaks in the way needed by this person now, and naturally exists in a form suitable for that space. Instead of the avatar's visuals, voice, and conversation playing separately, we will complete it as a perfect single personality to create an experience where people do not feel awkwardness.
As confirmed through the stories of the three people, Perso Interactive is moving beyond AI that merely answers inside a screen to Physical AI that provides practical help through natural 'conversations' everywhere people stay, such as exhibition halls, stores, vehicles, and online spaces. ESTsoft plans to continue expanding Perso Interactive to various countries and industries, making moments where people and AI naturally meet and interact into reality one by one.
In the next episode, we will cover the story of these three people who, although finding their own answers in different positions of planning, development, and business, are united under the single goal of 'creating an AI experience that people can naturally accept'.
![[ESTsoft] Signed MOU with K Monster Group to Expand Global Business of 'Perso AI'](https://framerusercontent.com/images/NfRLJQtbDUIdq8ujJl0TtG4S4.png?width=800&height=536)
![[EST-SELF] Different roles, but one answer: Perso Interactive Business Division connecting people and the world ②](https://framerusercontent.com/images/E8mP9HceNcsHILR1wFLKy1lBdyc.png?width=1280&height=720)
