JobTwine founder Vikrant Mahajan shares how an ai interview platform solves high-volume screening, prevents cheating, and acts as an ai interview copilot.
JayT
The Digital Twin
This piece is adapted from a conversation between Vikrant Mahajan, Founder and CEO of JobTwine, an end-to-end AI Recruiting Platform and Chris Russell, Managing Director of RecTech Media and host of RecTech: The Recruiting Technology Podcast.
I started JobTwine in early 2022, but the idea took shape years earlier. Before this, I was CTO at Doc ASAP, a health tech company acquired by Optum in 2021, and before that I led engineering for a product at Adcentive, which was also acquired by Optum, back in 2018. In both of those roles, I was hiring constantly while also trying to scale the business and ship product. What struck me was how painful and inconsistent interviewing was, no matter how much coaching or structure you tried to give a team. Around 2019 and 2020, AI note-taking and conversation-intelligence tools started showing up, and that's really what got me thinking about applying that same intelligence to the hiring conversation itself. Then ChatGPT launched in late 2022, and everything accelerated. Over the last two years, we've leaned heavily into voice agents as a result — and that's the foundation of the ai interview platform we've built at JobTwine.
Where JobTwine Stands Today
We at JobTwine have a couple of customers in the U.S. right now and a handful of pilots running with mid-size to large companies, that's intentionally our focus. We've raised about $3 million so far and we still consider ourselves pre-seed. With some new go-to-market hires joining the team, our goal over the next six to nine months is to take this to the next level.
But Why Mid-Size Companies, Not Startups?
It comes down to where the pain is most acute. Mid-size and larger organizations are the ones dealing with real volume,thousands of applicants, while also trying to keep their interview process consistent across a growing number of interviewers. Early-stage startups usually are not hiring at a scale where that inconsistency becomes a real cost yet. That's where we think we can have the most impact right now, particularly through automated candidate screening.
What's Changed Most in Recruiting
The sheer volume and polish of resumes coming in. That sounds like a good problem to have, but it actually makes screening harder, because you can't rely on what's on paper the way you used to. At the same time, recruiters have a fixed number of hours in a day. That mismatch is exactly why AI has become necessary rather than optional, it's simply not possible anymore to give every applicant a fair, thorough conversation using human bandwidth alone. True AI recruitment intelligence means bridging that distance between paper qualifications and real coding evidence.
The Pushback We Hear
Two things, almost every time. First is compliance — are the models biased, are candidates being assessed fairly and consistently. That's a legitimate concern, and it means companies need to vet vendors carefully to avoid systemic, unstructured processes that mask hiring bias. Second is more of an emotional reaction than a technical one — a worry that AI simply can't replicate what a skilled human interviewer does. That second concern is less about the reality of the technology and more about fear of the unknown.
Does an AI Interviewer Scare Away Top Talent?
I understand where that comes from, but I don't think it holds up in practice. The reality most recruiting teams are living with is that it's just not possible to give every single applicant a fair, in-depth conversation using human interviewers alone — there simply aren't enough hours in the day. AI lets you extend that fair shot to more candidates. There's also a scheduling benefit people overlook: a lot of frontline and healthcare candidates need to interview at 2am or on a weekend, and an ai interviewer can accommodate that in a way a traditional 9-to-5 recruiting team can't.
How Customizable Is the Candidate Experience?
Quite a bit. Companies can build an AI Avatar that looks and sounds like an actual recruiter, hiring manager, or even the CEO, so the experience feels like a natural extension of their culture rather than a generic automated tool. We also support around 40 languages, along with regional dialects and accents — for example, distinguishing Brazilian Portuguese from European Portuguese, or accounting for the many different dialects of Mandarin. Even speaking pace can be adjusted, since candidates in technical roles often prefer a faster-paced conversation.
What the Data Shows
On average, a skilled-role conversation runs 15 to 20 minutes, with 8 to 10 questions. Those aren't scripted, sequential questions either — there's real follow-up and probing based on how the candidate responds, which keeps the conversation feeling natural rather than robotic. For high-volume, lower-skill roles, even a 10-minute screening is usually enough to surface the signals that matter most.
Completion Rates
We see completion rates in the range of 60 to 65 percent. That's actually where the human recruiter still plays an important role — following up personally, sending reminders, and warming the candidate up to actually complete the process rather than letting the invitation go cold.
Handling Cheating and Fraud
Candidates cheating with AI during interviews seems like an obvious risk. It's a real concern, and we built fraud detection into the platform specifically because of it. We look at things like whether a candidate appears to switch tabs mid-interview, whether their eye movement suggests they're reading an answer off another screen, and whether the same person stays in frame the whole time. We also look at how an answer sounds — responses that come across as unusually smooth, with very few pauses or filler words, often indicate someone reading a scripted or AI-generated response rather than speaking naturally. When those signals line up, we generate a suspicion score so the recruiter knows to take a closer look.
Fitting Into the Recruiter's Workflow
Beyond the interview itself, JobTwine integrates directly with the ATS, so once an interview is complete, the transcript and scorecard flow right back into the system recruiters are already using. One feature people especially like is that it auto-populates the standard feedback forms most ATS platforms require — it reads the transcript and generates written answers to whatever competencies the form is asking about, so recruiters aren't manually writing up notes after every conversation. We also offer an ai interview copilot mode for live, human-led interviews, where the AI listens in on a call and nudges the interviewer in real time on what to ask or probe further, which helps keep a large team of interviewers consistent with what the hiring manager actually wants. For teams comparing tools or exploring Metaview alternatives, this dual capability — full AI-led interviews plus live copilot support — is what sets JobTwine apart.
What About Candidates Showing Up With Their Own AI Agents?
We've actually already run into early versions of this — proof-of-concept cases where a deepfake stood in for a real candidate. Right now, it's still limited, mostly because of lag and the complexity of generating a natural, real-time, probing conversation with zero delay. But the pace of progress in this space tells me that the gap won't stay open for long. It's actually a big part of why platforms like ours need to keep evolving, you need a system capable of recognizing when the person on the other end of the conversation isn't actually a live human.
Listen to the full podcast episode on RecTech Media: Watch on YouTube
