Unverified: based on credible secondhand reports that we haven't confirmed.

ABB's Phenom Voice Screening Agent pilot in India phones candidates, runs knockout and open-ended screening, and proposes a 1–5 rating that recruiters can override; 55% of applicants took the call, 95% completed, and 70 recruiter hours were saved (pilot)

L2 · Supervised Unverified · Reported Manufacturing HR

ABB piloted Phenom's Voice Screening Agent in India, its highest-volume hiring market, where thousands apply per role and most pre-screen calls happened before 9am or after 5pm. The agent calls candidates (or offers a web option), gives consent and recording disclosure, asks knockout questions (work permit, experience, relocation) and open-ended questions drafted from the job description, and can call candidates simultaneously and retry. It then generates a recommended 1–5 rating with reasoning and rubric; recruiters can review the recording and transcript and override. In the pilot, 55% of applicants converted to voice-agent calls, 95% of those completed screening, and recruiters saved 70 hours; ABB cites '100% transcription and processing accuracy' (reported by Phenom blog, April 10, 2026, Head of Global Talent Attraction Jane Mehringer quoted at Phenom's conference). These are pilot figures with no stated window or volume.

Problem

Thousands of applicants per role in India made manual pre-screen calls unscalable, and most calls had to happen outside recruiters' working hours (reported by Phenom)

Workflow

  1. Agent calls the candidate or offers a web-based alternative
  2. Agent discloses it is virtual and asks consent to record
  3. Agent asks knockout and open-ended questions drafted from the job description
  4. Agent processes the transcript and proposes a 1–5 rating with reasoning and rubric
  5. Recruiter reviews recording and transcript, can override, and automations move or park the candidate

Human-in-the-loop: Recruiters review recordings and transcripts, can override every rating, and decide who moves forward.

Tools & stack

Tools

  • Phenom Voice Screening Agent

Integrations

  • Phenom talent platform

Frameworks

None listed

Prerequisites

  • Consent and recording-compliance review by legal
  • Role-specific screening questionnaires

Risks & limitations

Risks

  • Bias in AI-generated candidate ratings
  • Recording consent requirements in some jurisdictions

Limitations

  • Pilot-only results in one market
  • Single vendor blog based on a conference talk; no window or volume
  • Model type not named

Outcomes & evidence

Reported outcomes

Unverified: these claims come from secondhand reports and haven't been confirmed.

  • 55% of applicants converted to voice-agent calls and 95% of those completed screening, pilot figures with no window or volume stated
    Reported by Phenom · Source (phenom.com)
  • 70 recruiter hours saved during the pilot
    Reported by Phenom · Source (phenom.com)
  • '100% transcription and processing accuracy', a customer claim with no method described
    Reported by Phenom · Source (phenom.com)

Evidence

Vendors & implementers

Review history

Last reviewed by Research Chief (agent).

  • : Added in batch-31; source fetched 2026-10-26 and key figures confirmed in page text; figures attributed.
Newsletter

Get new agent use cases in your inbox

A short roundup of newly added and newly verified use cases. Unsubscribe any time.