What we are asked before a deal

An AI recruiting platform for the first stage of hiring. Job creation, application intake, applicant review, the AI interview and the candidate report, all in one console. The AI proposes interview questions, a recruiter reviews and edits them, and the chosen candidates take an AI interview. Candidates answer by voice in a phone browser; the recruiter reads a Japanese report with a score and rationale per criterion, the quoted evidence in the original language (with a Japanese translation when the interview was in English), and the full transcript and audio.

No. Hitoshiru is designed never to decide an outcome. The AI proposes questions, conducts the interview, organises documents and answers, and produces a score per criterion. Each score carries the utterance behind it, and a recruiter can overwrite any score with a reason. Whom to meet and whom to hire remain a person's decisions, as they always were.

Yes. When you create a job you choose, from a fixed list, what the interview should establish: whether the experience holds up, how the person behaves at work, why this job and whether they will stay, a first-round screen, communication in the role, or potential. Up to two per job. The suggested criteria and questions, the interviewer's follow-ups and the report's summary are written for that choice, and the report says what the interview was set to establish. There is no free-text option and no culture-fit option, on purpose: those are where the topics the Ministry of Health, Labour and Welfare asks employers to keep out of hiring enter under another name.

Yes. For mid-career and graduate hiring alike, the interview runs in Japanese and the report arrives in Japanese. Questions are set per job to fit the role and the profile sought, so it serves first-round screening across sales, engineering, back-office and other roles. For candidates from overseas, a job can offer the interview in English.

No, and no account either. They open the link in a phone browser, check the microphone and the connection, answer a practice question, and begin. If the connection drops, the interview resumes from the last answered question.

It depends on the number of criteria and how much follow-up the answers need. Thin answers are followed up on the spot, so two candidates on the same job can take different lengths of time. An estimate is shown on screen once the criteria are built.

We do not predict attrition. We have no evidence that would justify claiming to judge in advance who will stay. What Hitoshiru gives you is the material for the judgement: in the AI interview, every candidate is asked, on the same terms, about the experience, the reasons for leaving the last role and the view of the working conditions that a document or a short call does not reach. The decision stays with a person.

Every follow-up question the AI writes during the interview is checked, before it reaches the candidate, by a separate AI model against items 1–11 of the Ministry of Health, Labour and Welfare's 14 topics that may lead to discrimination in hiring — registered domicile and birthplace, family, housing, living environment, beliefs and so on. Speech synthesis may start while the check runs, but the audio stays on the server until the question is approved. A match is discarded before the candidate hears it and the follow-up is written again; after two discards in a row, the interview moves to the job's next question. The discard is written to the audit log. The criteria and questions your team writes are screened for words touching the same eleven topics when they are saved. A flagged item can be saved, but the job cannot be published until someone records why it is there. Items 12–14 (background checks, application forms, medical examinations) concern how selection is run rather than what is asked, and no automated check covers them at present.

Yes. Speech recognition, response generation, speech synthesis and report writing use OpenAI's API. The data is handled by OpenAI, L.L.C., a company in the United States, so under the APPI this is a provision of personal data to a third party in a foreign country. Before the interview, the consent screen names the recipient, its country and its purpose, says that the handling includes provision to a third party outside Japan, and asks the candidate to tick a box to agree. The version of the consent wording and the provider list the candidate saw are stored with the consent record.

There is no fixed retention period. Data is deleted from the live system when you ask; encrypted database backups keep it until they age out, within about five weeks. No scheduled expiry is implemented, and we do not describe a mechanism we have not built as a policy. A retention period can be set by contract.

Not at present. Plans are discussed in conversation. Security questionnaires are answered individually.