Interviewer.AI

Company overview

Interviewer.AI is a Singapore-headquartered HR-tech vendor founded in 2018 by Sunny Saurabh (CEO), Srividya Gopani, and Manoj Salian (incorporated as UEN 201837827Z). Its core product is an end-to-end asynchronous video interview platform that lets employers, universities, and SMBs screen large applicant volumes through recorded video responses, with the system generating an 'AI Score' and a 'Deep AI Score Breakdown' from a combination of I/O-psychology heuristics and machine-learning models that score communication, response relevancy, and resume alignment. The suite also includes avatar-based conversational interviews, candidate mock interviews, and university admissions assessments. Pricing is published in self-serve mid-market tiers (roughly USD 53–67/month per plan) alongside an annual enterprise option, placing it in the mid price tier.

Regulatory exposure

As an automated tool that 'substantially assists' hiring shortlisting, Interviewer.AI sits squarely in scope of NYC Local Law 144 (AEDT), the EU AI Act's high-risk employment classification, and US state automated-decision rules in Illinois and Colorado. Its public materials carry no NYC LL 144 bias-audit summary and no independent (BABL/Warden/Holistic/ORCAA-type) audit, which is the single largest exposure given the product historically analyzes facial expression, eye-contact, energy, and 'emotional state' signals — a category increasingly restricted as prohibited emotion-recognition under the EU AI Act. Notably, its detailed privacy policy now states emotion recognition is 'not performed on any deployment' and that biometric/facial features are 'optional only,' which conflicts with the still-live public Explainable-AI page describing facial and emotional-state scoring; this inconsistency is itself a compliance risk. On the positive side, the privacy policy is unusually substantive: ISO/IEC 27001:2022 certification, a documented human-review-only decision model, GDPR/DPA/DPIA references, CCPA automated-decision opt-out, AI-vendor training-data exclusions, 24-hour raw-media deletion, and demographic-identifier separation before model evaluation.

Path to a higher score

The highest-impact move is to commission and publicly post an independent bias audit (NYC LL 144 disparate-impact summary plus an auditor such as BABL AI, Warden AI, or Holistic AI), ideally on a yearly cadence, since the vendor currently publishes none. Next, reconcile the contradiction between the Explainable-AI marketing page (facial/emotion/eye-contact analysis) and the privacy policy (emotion recognition disclaimed) by publishing a single authoritative system/model card and an Article 11-style technical documentation pack with a clear feature inventory and exclusion list. Adding EU AI Act deployer guidance (a FRIA/Article 27 template and instructions-for-use), pursuing ISO 42001, and standing up a public post-market monitoring surface (status page, model-update changelog, security/incident contact) would move the vendor from a 'privacy-policy-only' compliance posture toward verifiable, audit-backed transparency.