Scoring 95 in Math but Saying "I'm Unhappy"... AI That Reads Children's Hidden Feelings Better Than Teachers [AI Revolution]

Interview with Tebasoft CEO Oh Jeongseop
"Simspace" Adopted by 1,500 Schools
Analyzing Student Diaries to Detect 46 Emotional Signals

Gemini-generated image

"How about we write a short reflection on how this test went for you?"

After finishing a math unit test, an elementary school teacher asked the children to write a short piece. Since the class had been split into fast and slow learners, the teacher thought the outcome would be obvious. The expectation was that students who were good at math would say they were satisfied, while those who had struggled to keep up would write that it was hard.

However, the results analyzed by Tebasoft's artificial intelligence (AI) emotion analysis platform "Simseupeiseu" were different. Even among children who did well in math and scored 95 points, there were many signals of "unhappiness" and "stress." Conversely, there were also children whose grades were not good but whose emotional state was stable. Only then did the teacher realize:

"The reason I assumed some kids would be having a hard time, and the reason I brushed others off as fine, all came from an adult's standards."

Simseupeiseu, which analyzes the short texts written by students with AI to visualize their emotional states, brings to light signals in the classroom that are easy to miss. It shows the emotional temperature of the entire class at a glance, and visualizes emotional states such as stress, anxiety, and depression with data. On the 3rd, Oh Jeongseop, CEO of Tebasoft, who met with The Asia Business Daily (photo), said, "What surprises teachers the most is precisely this kind of reversal."

On the 3rd, Oh Jeongseop, CEO of Tebasoft, is introducing the Simseupeiseu feature in an interview with The Asia Business Daily. Photo by Park Yujin

After completing a PhD course in computer engineering and working at Samsung Thales and as a research professor at KAIST, CEO Oh founded Tebasoft in 2022. The core of the Simseupeiseu business model came from the problem awareness of his wife, who is an elementary school teacher. Starting from the on-site concern of "I don't understand what is going on in the children's hearts," they built a system in which AI analyzes students' diaries and presents the results on a dashboard that teachers can grasp at a glance.

In 2024, with the introduction of AI-powered digital textbooks, the school environment changed rapidly. Basic infrastructure such as Chromebooks was installed, and the use of edtech expanded at the same time. Amid these changes, Simseupeiseu began showing the emotional temperature of the entire class using a red-yellow-green traffic light, and quantifying stress, depression, and anxiety so that teachers could understand the situation at a glance. The data revealed patterns in which children who scored 95 points in math still felt unhappy, and students who repeatedly used expressions like "annoyed" eventually ended up in conflicts.

The nuances of emotions revealed in the classroom in this way were difficult to fully capture with conventional emotion analysis methods. Feeling the limitations of the initial 13 emotion categories, Simseupeiseu switched to a large language model (LLM)-based system and expanded its emotion framework to 46 categories. This allows teachers to view children's conditions not simply as abnormal versus normal, but as trends and flows of change.

This is also where it differentiates itself from competing services. Rather than stopping at analyzing an individual's emotional state, it focuses on reading emotional flows within the school and class groups. The "group emotion analysis" technology, which detects unusually sharp emotional changes or signals of isolation within a group, is a core patent held by Tebasoft.

The Simseupeiseu subscription fee is around 40,000 won per class per month, and it is currently in use at 1,500 schools. Among them, 800 schools adopted it at no cost to the schools through a contract with the Seoul Metropolitan Office of Education, and starting this year a contract with the Gwangju Metropolitan Office of Education will also begin.

Having been selected at the end of 2025 for the Ministry of SMEs and Startups' technology startup support program Global Track, Tebasoft's next target is overseas school markets. Simseupeiseu currently supports eight languages, including Korean, English, Japanese, Chinese, and Spanish, and pilot programs are underway in schools in Mongolia and Japan. Discussions on adoption through local partners are also continuing in countries such as Vietnam and Sri Lanka.

The approximately 260,000 emotion data points currently accumulated in Simseupeiseu are used not only to further advance the model, but also as emotional indicators that can be applied in schools. CEO Oh explained, "There are many moments when teachers confirm with data what they had only vaguely assumed," adding, "In a classroom environment where there are limits to how closely you can watch each and every child, our goal is to serve as an auxiliary tool."