AI Voice Agent for Candidate Qualification

An outbound voice AI system that calls job applicants, loads their CV in real-time, verifies identity, assesses motivators, and determines role fit — deployed for one of Japan's largest recruitment firms.

24/7

Outbound Calling

10,000+

Hours Saved

5x

Cheaper Than Human Recruiter

Sentry AI, in partnership with DynaFlow, is building AI-powered recruitment voice agents across the Asia region.

High-volume recruitment is slow and expensive. Qualifying 10,000 candidates manually takes months of recruiter time — each call averaging 10–15 minutes, each hire costing more than it should. Delays in outreach mean top candidates accept other offers before they're even contacted.

Our AI voice agent for recruitment eliminates that bottleneck. It calls 10,000+ candidates in days instead of months, at a rate roughly 5x cheaper than a human recruiter. Every call uses real-time live function calls to pull the candidate's CV mid-conversation, assess fit against the role, and — if they're not the right match — recommend alternative jobs to them live on the call.

The result: faster time-to-hire, dramatically lower cost-per-screen, and zero candidates lost to slow follow-up. The system is now live with one of the largest recruitment companies in Japan, processing candidate qualifications at scale across multiple job categories.

How the call flow works

Each outbound call follows a structured qualification pipeline:

1

CV Retrieval

Before the call connects, the agent triggers a function call to load the candidate's CV and the job listing they applied for, giving it full context for the conversation.

2

Identity Verification

The agent opens by confirming it's speaking with the right person — verifying name, contact details, and that they did apply for the role in question.

3

Job Confirmation

The agent confirms the specific role the candidate applied for, outlines key details about the position, and checks the candidate is still interested in proceeding.

4

Motivator Assessment

A series of qualifying questions assess what matters most to the candidate — remote vs. in-office preference, salary expectations, commute tolerance, career growth priorities, and work-life balance.

5

Fit Determination

Based on the candidate's responses and the role requirements, the agent reaches a conclusion on whether the job is a strong fit and communicates this clearly to the candidate.

6

Alternative Job Matching

If the role isn't the right fit, the system triggers an asynchronous query against the full job database to surface alternative positions that better match the candidate's motivators and experience — keeping them engaged in the pipeline rather than lost.

The problem we solved

High-volume recruitment teams face a set of compounding challenges when qualifying candidates manually:

  • Recruiter bandwidth — Each qualification call takes 10-15 minutes, creating a hard ceiling on how many candidates can be processed daily
  • Inconsistent screening — Different recruiters ask different questions and weight answers differently, leading to inconsistent qualification decisions
  • Candidate drop-off — Delays in outreach mean candidates accept other offers before they're even contacted
  • Lost pipeline value — When a candidate isn't right for one role, they're rarely matched to alternatives and simply fall out of the funnel
  • Scaling cost — Processing more applications means hiring more recruiters, with no improvement in per-call efficiency

Technical architecture

The system was engineered for reliability and natural conversation flow at scale:

  • Real-time CV loading — Function calling retrieves candidate data and job details before the call connects, so the agent has full context from the first word
  • Structured conversation engine — The call follows a defined pipeline (verify, confirm, assess, determine) while allowing natural conversational flow within each stage
  • Motivator scoring — Candidate responses are scored against role attributes to produce a quantified fit assessment, not just a subjective opinion
  • Async job matching — When fit is negative, an asynchronous query runs against the job database to suggest alternatives without blocking the call flow
  • ATS integration — Qualification outcomes, motivator scores, and call summaries are written back to the applicant tracking system automatically
  • Low-latency voice — Optimised for natural-sounding conversation with minimal response delay, critical for candidate experience

Why this approach works

Speed to Contact

Candidates are called within minutes of applying, not days. Faster outreach means fewer drop-offs and more engaged candidates.

Consistent Qualification

Every candidate goes through the same structured assessment. No variation between recruiters, no questions missed, no bias in scoring.

No Candidate Left Behind

The async job matching ensures candidates who aren't right for one role are immediately surfaced alternatives — turning a rejection into a new opportunity.

Recruiter Time Reclaimed

Human recruiters focus on high-value activities — final interviews, client relationships, and closing — while the AI handles initial qualification at scale.

Scale Without Limits

Process thousands of applications without proportionally growing the recruitment team. The system scales with volume, not headcount.

Get in touch

If your recruitment team is spending too much time on initial candidate qualification calls, we can build something similar for your workflow. Every implementation is tailored to your specific screening criteria, ATS, and job categories.

More from this sector