Candidate-facing AI interview transparency layer

The moment a candidate realizes they have a problem is mid-interview, when the AI asks a follow-up that feels disconnected from what they just said, or immediately after, when they receive a rejection with no explanation of what criteria were used or where they fell short. They interacted with a black box and have no idea how to improve — or whether the evaluation was fair.

This gap persists for a structural reason: the buyer of AI interviewing software is an HR team or recruiter, not the candidate. The recruiter's complaint is about setup time and configuration. The candidate's complaint — 'I don't know how I'm being evaluated, which feels unfair' — has no economic voice in the purchasing decision. So vendors optimize for recruiter workflow, not candidate experience. Nobody is losing deals because candidates felt the process was opaque.

What the complaints make clear: candidates sometimes don't even know they're talking to an AI ('the system should clearly explain to users that they are interacting with AI at the start'); the AI scores borderline answers inconsistently with no explanation; and personal feedback based on the actual interview is absent. A candidate who freezes up because the interface feels robotic, or who doesn't understand why they failed, is a liability — they leave negative Glassdoor reviews, they tell their network, and they create legal exposure if the scoring can't be explained.

This is a business because every company using AI interviews faces this problem for every candidate who goes through the process, and the frequency scales directly with hiring volume. A company running 500 AI interviews a month has 500 candidates per month who experienced the black box. The buyer — HR directors and talent acquisition leads — increasingly has legal and reputational pressure to demonstrate fair, explainable hiring processes, especially as AI-in-hiring regulation tightens in the US and EU. That pressure recurs every hiring cycle.

What to build

Build a candidate-facing debrief layer that sits on top of AI interview platforms and automatically generates a plain-language post-interview summary for each candidate — showing which competencies were assessed, how their answers were interpreted, and where they can improve — without exposing proprietary scoring weights.

Where to start

Start with companies in highly regulated industries (financial services, healthcare) where explainability in hiring is already a compliance conversation — they have a named internal champion (legal or compliance) who will push HR to adopt this alongside whatever AI interview tool they're using.

The hard part

Getting API access to interview transcripts and scoring data from incumbent AI interview vendors is the core dependency — vendors may see this as competitive and restrict access, forcing a pivot to companies willing to export raw transcripts manually.

How it makes money

Per-interview fee charged to the employer (not the candidate) — something like $1–$3 per completed interview — so cost scales with hiring volume and the value is felt immediately in reduced candidate complaints and Glassdoor fallout.