An AI voice agent for recruitment that screened 3,000 applicants, and saved 300+ from the reject pile
DynaFlow gets more applicants than its recruiters can ever call back. We put an AI voice agent for recruitment on it, so every applicant gets reached, qualified, and, if they don't fit, matched to another job instead of ghosted.
3,000
Applicants called in 3 months
300+
Referred to other roles
5x
Cheaper than a recruiter
TL;DR
DynaFlow, a Japanese recruitment firm, gets far more applicants than its recruiters can call back, so good candidates get ghosted and quietly go elsewhere. We built an outbound voice agent that calls applicants in Japanese, pulls their CV from the LinkedIn job post in their Bullhorn ATS, and qualifies them. Over three months it made 3,000 calls, 1,800 answered first time, qualified 220 to the next stage, and, the part that mattered most, referred 300+ candidates who were not right for their role to other open jobs in the system. All for roughly a fifth of what a recruiter costs.
What is an AI voice agent for recruitment?
An AI voice agent for recruitment is a system that phones your applicants automatically, holds a real spoken conversation, and screens and qualifies them against the role, then writes everything back to your ATS. Think of it as voice AI recruitment software that handles first-pass calling at a scale no team of recruiters can match: every applicant reached within minutes, day or night, in whatever language your market speaks.
A good one does more than read a script. It pulls the candidate's CV before it dials, runs a branching qualification conversation, moves the strong matches to the next stage, and, the part most teams miss, refers the people who don't fit to your other open roles instead of letting them vanish. It is the same voice AI recruiter capability whether you think of it as candidate pre-screening, an AI voice agent for recruiting, or plain recruitment automation.
Agents like this one are registered, monitored and governed from Sentry's Agent Management Platform, and it fits inside the wider picture of how recruitment teams are using voice AI. Here is exactly what it looked like for one agency.
In Japanese recruitment, the bottleneck is not finding applicants, it is calling them all back. A single job post can pull in more candidates than a recruiter can realistically phone, so plenty of good people apply, hear nothing, and move on.
DynaFlow lives that problem at scale. Recruiters were spending their days on first-pass screening calls, and still could not reach everyone who applied.
Why do good candidates slip through?
Because there are simply too many to call by hand. Every applicant who is not phoned back quickly is a candidate who drifts to another agency. And the ones who do not fit the exact role they applied for tend to vanish entirely, even when there is another open job they would be perfect for.
Doing that outreach properly, calling every applicant, fast, in Japanese, would mean hiring more recruiters. That is expensive, and it still does not scale.
What does the voice agent do on a call?
It calls every candidate who applies to a job post, in Japanese, promptly after they apply. It pulls their CV straight from the LinkedIn job post, stored in DynaFlow's Bullhorn ATS, so it already knows who it is talking to.
From there it runs a branching qualification conversation against the role, moves the strong matches to the next stage, and keeps every answer structured back in the ATS.
How did it surface 300+ extra candidates?
This was the part that mattered most. When a candidate was not right for the role they applied to, the agent did not just thank them and hang up. It checked them against the other open jobs in DynaFlow's system and referred them on.
Across the three months, that turned 300+ would-be rejections into live referrals for other roles, candidates DynaFlow would otherwise have lost entirely.
What was the return?
In three months the agent made 3,000 calls, reached 1,800 candidates on the first attempt, qualified 220 to the next stage, and referred 300+ to other roles, all in Japanese.
| Applicants called | 3,000 (over 3 months) |
| Answered on first attempt | 1,800 (~60%) |
| Qualified to the next stage | 220 |
| Referred to other roles in the system | 300+ |
| Language | Japanese |
| Cost vs a recruiter | ~5x cheaper |
| Live in | 2 months |
On cost, the comparison is stark. A recruiter on an average Japanese salary, around ¥4.5 million a year, doing this outreach by hand costs roughly five times what the voice agent does it for, and a recruiter still cannot call every applicant the moment they apply.
The result
Every applicant called, in Japanese, the moment they applied, 220 qualified to the next stage and 300+ candidates rescued for other roles, all for around a fifth of what a recruiter costs.
Could this be your agency?
If you get more applicants than you can call back, lose good candidates to ghosting, or spend your recruiters' days on first-pass screening, this is exactly the kind of problem we solve. We start with a free AI Opportunity Audit: enter your company, connect the tools you already pay for, and see your three biggest AI opportunities mapped before you spend a dollar.
Frequently asked questions
What is an AI voice agent for recruitment?
An AI voice agent for recruitment automatically calls your applicants, holds a natural spoken conversation, and screens and qualifies them against the role before writing the outcome back to your ATS. It does first-pass candidate calling at a scale and speed no recruiting team can match: every applicant reached within minutes, day or night, in your market's language.
How does a voice AI recruiter work?
It pulls the candidate's CV from your ATS before it dials, runs a branching qualification conversation against the role, moves strong matches to the next stage, and refers the rest to your other open jobs. Everything is logged back into the systems you already run, and recruiters still make the final call.
How much does an AI voice agent for recruitment cost?
It varies with call volume and how deep the ATS and qualification logic go, but in production it typically runs around five times cheaper than having recruiters do the same first-pass outreach by hand, and unlike a recruiter it can call every applicant the moment they apply.
Won't candidates think it is a scam call?
The agent identifies itself and DynaFlow up front, and it calls promptly after someone applies, so it lands as a fast follow-up to the candidate's own application, not a cold call out of nowhere. In recruitment the bigger risk is the opposite: applicants who never get called at all and quietly go elsewhere.
What about candidates who normally get ghosted?
That is exactly what this fixes. Every applicant gets called, fast. And the ones who are not right for the role they applied to are referred to other open jobs in the system instead of disappearing into a reject pile.
How accurate is the qualification?
The agent qualifies against the role using the candidate's actual CV, pulled from the LinkedIn job post and stored in their Bullhorn ATS. It moves the right people to the next stage and refers the rest; recruiters still make the final call.
Does it really run in Japanese?
Yes. The entire conversation, qualification and referral happens in natural Japanese.
Where does the candidate data come from, and is it secure?
It reads from the systems DynaFlow already runs, the LinkedIn job post and their Bullhorn ATS, rather than moving candidate data somewhere new.
More case studies
Jade Financial: AI Voice Receptionist for a Finance Firm
A custom voice agent handles 30-50 inbound calls a day, reading from a new cloud database in real time, so the team only takes the calls that matter.
Internal Voice AI for Real Estate
An internal AI voice agent platform built for US real estate operations.
Healthcare Voice AI Across 15 Practices
A conversational assistant handling routine patient calls for a medical network of 180,000 patients, with data that never leaves the region.
MacroActive: An AI SDR That Ramps in Days
An AI operating system for a sales team, so reps get good faster and leads never go cold waiting for a callback.