reddit · SideProject · 10-02 03:50 62 分 ← 返回列表

I built an AI interviewer, here are 5 things I learned watching recruiters actually use it

Context: I built ProofRound, an AI interviewer that turns a job description plus a candidate's CV into a tailored interview kit in under 20 seconds. Questions map to the actual claims in the CV, and the output is a structured evaluation a hiring manager can defend. I spend a lot of time watching recruiters use it. Some things surprised me: Nobody trusts a score without receipts. Early version gave a rating per answer. Everyone ignored it. The moment we made every question cite the spec

五维评分

付费意愿

需求真实度

个人可落地

竞争空间

新鲜度

AI 分析

痛点：招聘官和 hiring manager 要花大量时间手工把 JD 和简历整理成定制化面试题库与评估表，且缺乏可辩护的结构化评估依据。

目标用户：招聘官、HR、中小企业 hiring manager 和面试负责人

证据：The feedback we get most is some version of "I used to spend my Sunday night building these". / Recruiters don't want harder questions, they want fairer ones.

变现：B2B SaaS 订阅：按招聘席位或按面试场次收费，可加高级评估报告、团队协作与 ATS 集成做增值层

MVP：输入 JD + 简历，20 秒内生成结构化面试题与评分表，每道题都引用简历中的具体表述作为考察依据，输出可导出的候选人评估表

竞品：HireVue、Interviewer.AI、MetaView、BrightHire、Guide

摘要：开发者发布 AI 面试官 ProofRound，20 秒内把 JD 加简历生成定制化面试题库和结构化评估，并总结了招聘官实际使用中的 5 条经验。最有价值的信号是招聘官反馈『我以前周日晚上都在做这些』，说明手工出面试题是真实耗时的痛点；同时 AI 润色简历普及后，面试成为唯一真实信息来源。但该方向竞品密集，帖子本身是创始人自述，未见明确付费数据。