Hard Kr Exclusive | An AI Education Enterprise Secures Nearly RMB 10 Million in Angel Round Financing, Providing a One-Stop AI Exam Preparation Assistant for 40 Million College Students Nationwide
Yingke has learned that Guangzhou Tiyuan Network Technology Co., Ltd. (hereinafter referred to as "Tiyuan Technology") recently completed an angel - round financing of nearly ten million yuan. The investor is Hongtai Smart Valley. The funds from this round will be mainly used for market promotion and continuous technological R & D of the education agent platform.
Tiyuan Technology was founded in 2025 and is committed to becoming a one - stop AI assistant in the field of college students' exam preparation. The company uses large AI models to drive personalized learning plans, covering the entire process of courses, question - brushing, question - answering, information, application, etc. Its service scope covers multiple tracks such as postgraduate entrance exams, civil service exams, upgrading from junior college to undergraduate, CET - 4 and CET - 6, IELTS and TOEFL, as well as teacher qualification, accounting, judicial, and medical exams.
The founder, Wu Qiuheng, has been deeply involved in the college students' education track for 15 years. He started with certificate training during his college years and later founded an offline education and training company with an annual revenue of over 100 million yuan. In 2023, after the previous company was successfully acquired, he saw the opportunity for AI to enter the education scenario.
Wu Qiuheng believes that 2026 will usher in a dividend period for the explosion of AI education applications nationwide. Currently, although the market for college students' exam preparation has strong demand, it still faces the dilemma of the "impossible triangle" in traditional education and training, that is, it is impossible to achieve large - scale, personalized, and high - quality education simultaneously. Traditional offline education and training is often very expensive, costing tens of thousands of yuan. Fragmented question - answering is difficult to get an immediate response, and the problem - solving accuracy of general large models in the current market is often not satisfactory. In response to these pain points, Tiyuan Technology has independently developed a vertical large model for higher education, integrating more than 10 million question data, more than 100,000 structured knowledge points, and multi - modal video slices. The problem - solving accuracy has been improved to over 95%.
Its core product matrix includes AI digital human micro - classes, AI intelligent analysis reports, knowledge flashcards, etc. Among them, the AI digital human micro - class can provide personalized explanations "24/7". When students encounter difficulties while brushing questions, the system can instantly generate exclusive digital human explanation videos. Through the "AI + real - person teaching and research" model, the cost of traditional training courses worth tens of thousands of yuan is compressed to a subscription fee of a few hundred yuan, greatly reducing the threshold for students' exam preparation.
The "AI intelligent analysis report" is customized for free throughout the process, focusing on students' core needs for further education. Firstly, it determines the direction. Based on students' grades, subject advantages, and interests, it prioritizes sorting out the scope of suitable universities and corresponding major directions, clarifying the boundaries of further education choices. Secondly, it analyzes the admission scores. It synchronously marks the admission score data of target universities in recent years, analyzes the admission probability based on students' grades, and divides the universities into three gradients: challenging, stable, and conservative. Finally, it can generate a targeted study plan to help students strengthen their weak points and improve their strengths according to their further education goals.
In terms of team configuration, the company has both educational and technological genes. In addition to Wu Qiuheng, who is well - versed in the campus market, the technical director has many years of experience as a middle - and high - level executive at Alibaba, with rich experience and comprehensive technology. The teaching and research director has led the design of the curriculum system for millions of students and has taught on the front line. Currently, Tiyuan Technology has completed the initial product R & D and entered the internal testing stage, with rapid user growth.
In terms of business model, in addition to the subscription system for C - end students, Tiyuan also arranges B - end course and system SaaS services and product OEM services. It is open to cooperation with universities across the country and various education and training companies of all sizes, providing technological empowerment for education institutions and hardware manufacturers. Wu Qiuheng said that the company expects to achieve a revenue of 25 million yuan in 2026 and exceed 200 million yuan in 2028.
We had a chat with Wu Qiuheng about his entrepreneurial motivation and plans. The following is an excerpt from the interview:
Yingke: You have been deeply involved in the college students' education track for many years, and your previous company was also acquired. Why did you choose to enter the market with AI this time?
Wu Qiuheng: Since my college days, I have been engaged in various certificate training such as computer level 2 and teacher qualification exams around the college market. My entrepreneurship after graduation has never left college students. In the next 5 - 10 years, AI is a new wave in China and even the world. In such a great era, how to combine AI with college education and training and make it land is a question we have been thinking about. This time when starting a new business, the one - stop AI assistant for exam preparation we developed still targets the 40 million college students in China and young people who are interested in further studies or obtaining professional qualification certificates. Our in - depth understanding of this group is our core advantage.
The biggest pain points of traditional education and training are efficiency and cost. A teacher can only serve dozens or hundreds of students. However, AI can infinitely improve efficiency and at the same time reduce the cost to an extremely low level. AI is the optimal solution to achieve "teaching students according to their aptitude". We don't just apply open - source large models. Instead, we build an "education agent" that really understands exams and students through our self - developed vertical education model and a large amount of original high - quality data sets.
Yingke: Compared with the question - searching software or recorded courses on the market, what are the competitive advantages of Tiyuan Technology?
Wu Qiuheng: Traditional question - brushing software only provides answers and fixed explanations, without solving the problems of "understanding" and personalization. Recorded courses are just traditional one - way outputs and cannot answer questions about specific knowledge points in a targeted way. Our AI digital human is real - time. If you don't understand a question, the system will instantly generate a corresponding explanation video and answer questions in real - time, just like having a private tutor sitting beside you. Moreover, the problem - solving accuracy of our system reached over 95% in internal tests, which is difficult for general large models to achieve.
Yingke: The college student market is relatively closed. What mature strategies does Tiyuan have for market promotion?
Wu Qiuheng: The closed nature of the campus market is actually our advantage. Because it is completely different from the logic of public - domain advertising in other markets. The ROI of public - domain advertising is often very low, and the advertising cost is high. We are better at using the fission of social and entertainment content. We will use our self - built campus vertical media, link with college student bloggers and influencers across the network for product promotion, and our self - built more than 10,000 communities to make students become promoters themselves. This kind of communication based on acquaintance relationships in a closed circle is very fast and more efficient. There are 40 million college students every year, which is a huge market.
Yingke: As a start - up company, how do you build your data barrier?
Wu Qiuheng: Data is our most core moat. We have accumulated more than 10 million question data and more than 100,000 knowledge point maps. More importantly, we have the real learning behavior data of 1 million college students. Through user feedback and error correction, we can conduct continuous supervised fine - tuning. Once the user data starts to circulate, it is difficult for latecomers to catch up with the fineness of our model in a short time.