[ROXX] Development and Full-Scale Implementation of 'AI Feedback System' to Improve Interview Quality Begins

ROXX Inc. (Headquarters: Shinjuku-ku, Tokyo; Representative Director and President: Taro Nakajima; TSE Growth: 241A; hereinafter ROXX) has independently developed an 'AI Feedback System' that uses AI to improve the quality of business negotiations and interviews, and has begun full-scale implementation for its in-house career advisors (hereinafter CA) on the job change platform 'Z Career'. Concurrently, ROXX has also initiated full-scale development with a view to external provision, including for sales professionals.

This system enables immediate comparison and analysis of interview content with high-performer data, and seamlessly executes comprehensive analysis, evaluation, and feedback. This transforms the 'tacit knowledge' of high-performers, which was traditionally implicit, into 'explicit knowledge' and standardizes the coaching process that previously relied on managers' experience. This establishes a system where all CAs can autonomously and continuously improve their interview quality.

Background

Since its establishment, ROXX has emphasized the 'unique value of human intervention' by CAs, providing support tailored to each job seeker. We believe that the value of an excellent CA in recruitment is not in mechanically matching conditions, but in drawing out what job seekers 'truly value' through dialogue, which they may not yet be able to articulate themselves, and supporting their decision to take a step forward in choosing a career that will occupy one-third of their lives.

However, in reviewing interviews and training CAs, organizational challenges arose: reviewing interview logs and providing guidance was time-consuming, detailed individual feedback was difficult, and the accumulated expertise of high-performers could not be fully reflected in training. In addition to the resource shortage on the part of trainers, there was a question of how to formalize the 'tacit knowledge,' which depended on individual experience and intuition, into 'explicit knowledge.'

Based on the idea of 'automating routine tasks through AI utilization and allowing humans to concentrate on more essential and creative work,' ROXX applied this concept to the recruitment field. By structuring the expertise of high-performers and having AI learn it, a system was built where AI can instantly analyze, evaluate, and provide feedback on interview content after each interview conducted by a CA. This allows CAs, after conducting interviews with job seekers, to review the differences in interview methods compared to high-performers, identify good points and areas for improvement, and understand specific ways to bridge these gaps. This enables them to concretely grasp improvement points and apply them to their self-reflection.

Key Functions of the 'AI Feedback System'

1. Interview scoring that quantitatively visualizes interview quality AI automatically analyzes interview logs, compares them with pre-set benchmarks of high-performers, and evaluates 'CA-job seeker speaking ratio,' 'questioning techniques,' and 'coverage of important themes to be heard,' calculating a score. This allows CAs to objectively review data such as 'why this interview went well' or 'where an opportunity was missed,' transforming the 'feel for the interview' that previously relied on intuition into a verifiable and comparable metric for everyone.

2. Qualitative feedback comments compared to high-performer interviews Since an 'ideal interview model' is incorporated into the AI beforehand, immediately after an interview, the AI comments on specific good points and areas for improvement for each item, such as 'active listening and empathy' and 'proposal skills.' It automatically generates specific, situation-appropriate feedback like, 'In this situation, you could have potentially approached the core of the job change motivation by further probing the job seeker's past original experiences with a question like XX.' This means that 'insights' that managers traditionally took several hours to review logs and convey orally can now reach CAs within minutes of the interview. This shifts the CA's growth cycle from 'waiting to be taught' to 'reflecting immediately and applying to the next situation.'

3. Optimization of trends and countermeasures in human resource management through filtering by problematic items and comparative analysis The system can filter by problematic items that have not met the benchmark and extract reports for comparative analysis by individual CA. This allows for a structural understanding of trends within the team. Consequently, management challenges such as identifying which issues are most significant, what kind of training should be conducted, and where business processes should be improved become clear, enabling the definition of specific countermeasures. Furthermore, the AI can instantly organize and summarize information necessary for the next interview or job recommendation, outputting structured information such as the job seeker's career axis, concerns, and key statements that influenced their decision-making. For example, this can significantly reduce the time required for member handovers and preparation for the next interview.

Future Developments

Moving forward, in addition to 'AI role-playing' to accelerate the early readiness of CAs.