AI Voice Agents for Recruitment and Candidate Outreach
Recruitment runs on speed and volume at the same time, which is why the follow-up is always the thing that slips. Zemora agents work candidate lists by phone and text, screen on the handful of facts that actually decide a shortlist, and give recruiters back the conversations worth having.
- Consistent screening across hundreds of candidates
- Phone and SMS on one candidate record
- Transcript on every screening conversation
- Warm transfer to the recruiter who owns the role
Two hundred applicants, one recruiter, a Tuesday
A role goes live and two hundred applications arrive in a week. The recruiter working it has four other roles open. What happens next is not a process failure so much as arithmetic: the first thirty applicants get a proper call, the next seventy get a text, and the rest get nothing – including, statistically, some of the best ones.
The other half of the problem is that candidates are being recruited by everyone at once. A candidate who applies on Monday and is contacted on Thursday has already had three other conversations. In recruitment, speed of first contact is not a nicety, it is the whole competitive position.
The five facts that decide a shortlist
Not a competency interview – the disqualifying facts, established consistently.
Availability
Notice period, start date and whether they are actively looking or browsing. The single most common reason a promising candidate wastes a week of process.
Location and travel
Where they are, what commute they will accept, and whether the role’s on-site pattern works. Cheap to establish, expensive to discover at offer stage.
Rate or salary expectation
Asked early, because a gap between expectation and budget does not close later – it just surfaces after everyone has invested four hours.
Right to work and certification
Whether they hold what the role legally requires. The agent captures what the candidate states; verification stays a human step.
Genuine interest in this role
Whether they recognise the role and want it, as opposed to having applied to forty things in one evening.
The same conversation on candidate two hundred
The argument for automating screening is not that it is faster, though it is. It is that it is consistent. A recruiter on their eleventh call of the day asks shorter questions than they did on their second – not through carelessness, but because that is what humans do. The agent asks candidate two hundred exactly what it asked candidate one.
That consistency is what makes the output comparable. Ten recruiters produce ten interpretations of “available soon”. One agent produces one, which is what turns screening notes into something you can actually sort a shortlist by.
- Same screening questions and same standard across the whole list
- Answers captured as structured fields, not free-text notes
- Concurrency capped per campaign so callbacks stay manageable
- Attempt limits and retry spacing set per campaign
Candidates answer texts
Candidates in employment cannot take a recruitment call at eleven on a Tuesday, which is why a phone-only process quietly filters for people who are not currently working. SMS reaches the passive candidates that voice alone misses, on the same record, so a candidate who ignores a call and replies to a text two hours later is one conversation.
Interview scheduling runs through appointment setting – including the reschedules, which in recruitment are constant.
What the recruiter picks up
A shortlisted candidate arrives with the transcript, the structured answers to each screening question, and the reason they cleared. The recruiter opens the conversation knowing notice period, rate expectation and location rather than spending the first ten minutes re-establishing them.
Where a candidate should speak to a person immediately – a strong fit for a role closing this week, or a question the agent should not answer – the call is transferred warm to the recruiter who owns the role, with the reason recorded.
What these agents do not decide
Worth being explicit, because screening automation attracts fair scrutiny. These agents collect and structure what a candidate tells them. They do not score candidates on protected characteristics, do not make hiring decisions, and do not replace the recruiter’s judgement about who is right for a role.
Right-to-work and certification claims are captured as stated by the candidate and remain a human verification step. Every screening conversation is retained as a transcript you can review, which is a stronger audit position than notes typed from memory after the call.
Recruitment questions
Does this replace a recruiter?
No. It handles first-contact screening at volume - availability, location, rate, right to work as stated, and genuine interest - and hands the recruiter a shortlist with transcripts. Judgement about fit stays with the recruiter.
Does it make hiring decisions or score candidates?
No. It collects and structures what the candidate says against the criteria you defined. It does not rank people on protected characteristics and it does not decide outcomes.
Can it reach passive candidates who will not take a call?
That is what the SMS agent is for. Voice and text share one candidate record, so a call ignored and a text answered are the same conversation.
Can it book interviews?
Yes, where booking is enabled for that agent - including handling the reschedules. See appointment setting.
How do you handle candidate data?
Encrypted in transit and at rest, retained on a schedule you set - 30, 60, 90 days or 12 months, or deleted on demand. Details are on the security overview.
Can different roles use different screening questions?
Yes. Scripts and criteria live on the campaign, so a warehouse role and a senior engineering role are separate campaigns with separate questions.
Screen a live role with an agent
Bring one open role and its applicant list. We will configure the screening questions you already ask and run it as a pilot.