Beijing-based education online platform Knowbox defines itself as AI education company, renaming the business after a USD 150 million funding round.
The Alibaba and Baidu-backed company held a press conference on July 18, announcing its (Chinese) brand name changed to Xiaohe Keji (小盒科技), literally "small box technology", from Zuoye Hezi (作业盒子), meaning "homework box".
"Knowbox will not only focus on homework thereafter," said Liu Ye (刘夜), the founder and CEO. After the name changing, the company is to target AI education, building an intelligent education ecosystem on-campus teaching and home counseling.
"AI education companies are essentially education companies at the current stage; It's not possible for AI teachers to replace human teachers," according to Xia Zhijin (夏志进), general partner at Vertex Capital.
Based on its website information, Knowbox is offering Chinese largest series products of teaching tools for public school teachers and students as well as AI-enabled adaptive learning services for students. The company also offers programming courses to children, laying out STEAM education.
Previously, the Edtech company focuses on optical character recognition (OCR), branding itself as a homework product, with a slogan "use Knowbox to take a picture and get homework corrected."
"Knowbox gained a good reputation for OCR correction. Teachers from public schools are using it, but it is not the only player to get into China's school system," said Gouyu (苟瑜), analyst from EO Intelligence.
Starting with mathematics education, Knowbox has expanded its services into 31 provinces, more than 400 cities, 100,000 schools, reached out to 40 million users and collected more than 40 billion pieces of studying behavior data. In OCR homework correction area, top player Zuoyebang (作业帮) has a user number upward of 200 million, and Xuebajun's (学霸君) user number achieves 90 million.
"Granularity is the key to adaptive learning," according to Gou Yu. It is easy to evaluate whether a student understands it if there is only one knowledge point, but normal teaching and homework questions involve multiple knowledge points. If a student fails to answer one question, AI technology helps to determine which knowledge piece is missed. "This is a process requiring long-term iterative algorithm development based on user data," she said.