The written test for the national civil service exam in 2025 has concluded, and candidates are about to enter the intense phase of interview preparation. Industry leader FENBI (02469) announced the launch of a premium interview AI evaluation product based on the company's self-developed domain-specific large model, officially released on December 13, with users able to experience it for a limited time at a price of 1 yuan per session.

The premium interview AI evaluation is the third AI tool launched by FENBI in this sector and is the first AI product specifically targeting interviews in the civil service examination industry. Following the subjects of essay writing and administrative reasoning, it further extends the service scope, providing support for the entire preparation process for users. In the second half of this year, FENBI has continued to make progress in the "AI + Professional Education" field, constantly strengthening its technical advantages and consolidating its leading position in the industry.

At the same time, the premium interview AI evaluation has been upgraded based on the original foundation, increasing evaluation efficiency by 5 times, providing users with efficient, professional, and detailed optimization suggestions to help them accurately identify problems and quickly improve their interview skills. It is understood that FENBI has previously applied AI to assist interviewers in their daily evaluations and Q&A work, significantly enhancing operational efficiency and effectively optimizing costs.

The difficulty of civil servant interviews lies not only in assessing candidates' knowledge accumulation and logical thinking but also in evaluating their ability to respond in real-time. With limited interview time, candidates need to overcome nervousness in a short period, quickly understand and analyze questions, and ensure they complete their answers within the allotted time.

Therefore, interview evaluation is a necessity for users during their preparation phase. Analysis by FENBI's Big Data shows that users who systematically study interview courses and undergo more than 30 mock training sessions and professional evaluations are more likely to achieve ideal results.

FENBI AI Interview Scenarios

FENBI introduces high-quality interview AI feedback through one-on-one simulations of real interview scenarios, guided by AI teachers throughout the answering process, creating an immersive interview experience that helps users quickly adapt to the exam environment and improve their stress resistance. At the same time, it provides real-time, multi-dimensional, professional feedback on users' answering videos, including answering time, topic relevance, answering framework, etc., and provides AI model answers. It offers systematic guidance on individual strengths, weaknesses, and suggestions, helping users further optimize their thoughts.

Based on user responses, evaluation results are generated in real-time.

High-quality interview AI feedback covers six major frequently tested structured question types, based on over a million answering data accumulated by FENBI, supported by self-developed vertical large model technology, and continued participation of experienced teachers with ten years of teaching experience in training models. Through deep learning and multi-task fine-tuning, it has achieved a high degree of adaptation to the public examination interview process. By precisely identifying different answering steps of question types, high-quality interview AI feedback comprehensively assesses users' performance in language expression, logical structure, and depth of thought, and quantifies the advantages and disadvantages of answers through FENBI's internal scoring model.

FENBI disclosed in its Earnings Reports that AI applications have significantly improved the company's Operation efficiency, resulting in an increase in profitability. In the first half of 2024, FENBI's Net income was 0.278 billion yuan, a simultaneous increase of 240.9%, and adjusted Net income was 0.349 billion yuan, a year-on-year growth of 21.2%. A recent Research Report by Huaxi pointed out that in recent years, FENBI has made significant progress in the field of large language models (LLM), independently developing the first vertical large model focused on the public examination industry, and launched the 'AI Smart Teacher', providing reliable, accurate, and targeted tutoring services, saving on tutoring costs, and improving gross margin. With the increase in offline class occupancy rates and the teacher-student ratio of online high-quality classes under the OMO model, there is potential for further improvement in net margin.