Flighting Job Seeker Fraud

High-volume hiring used to be “hard.” Now it’s a knife fight.

Applicant fraud is out of control. Screening has become a full-time job no one asked for. And trying to scale sourcing at volume without burning your TA team into ash is practically impossible with legacy workflows.

So, we brought in the operators who actually had to survive this mess.

Rippling. Phantom. Foley. DoorDash. Lyft. Instacart.Real leaders breaking down what actually happened inside their funnels — the pain, the missteps, the “oh crap” moments — and the eventual breakthrough.

This is where Covey enters the story. AI-native, fraud-aware, screening-crushing, capacity-releasing — giving recruiters their time back so outbound sourcing can actually happen again.chad

Speakers:

• Jay Patel — Sr Director Talent, Rippling (ex DoorDash)

• Laura Stapleton — VP People, Foley (ex Engine)

• Derrick Gellidon — Head of Talent, Phantom (ex Lyft, Instacart)

• Vijay Mani — Co-Founder & CEO, Covey

• Featuring Chad & Cheese and a bourbon fueled discussion

If you care about volume hiring reality — not fantasy decks — this is the episode you don’t skip.

PODCAST TRANSCRIPTION (cometh)

0:00:00.2 Joel Cheesman: Is this a passing fad or is this here to stay, this AI fraud uh, phenomenon?

0:00:06.6 Derrick Gellidon: It... It's here to stay.

0:00:08.0 Laura Stapleton: Here to stay.

0:00:09.2 Jay Patel: Here to stay here.

0:00:10.7 Joel Cheesman: Here to... I know you say here to stay. Yeah. Like Chad and I do as well.

0:00:12.9 Derrick Gellidon: I have a slight vested interest.

[laughter]

0:00:19.2 Joel Cheesman: All right, guys, let's do this. Uh, we are the Chad and Cheese Podcast. Let's give it up for our panel here. I will introduce them quickly because we have a lot to cover. Uh, Derrick, the Don Gellidon.

0:00:30.6 Derrick Gellidon: Hello, everybody. Good to meet you all.

0:00:32.2 Joel Cheesman: He's head of talent at Phantom. He's formerly at Lyft and Instacart. We had to throw those name brands in there. Laura Stapleton, VP of People at Foley. She's also the ex-VP of People at Engine. Jay Patel. Little company called Rippling. Maybe you've heard of them. Uh, he's the senior talent director there. He was also at Doordash before that. My co-host, Chad Sowash is in the house. Feel free...

0:00:57.1 Chad Sowash: Give it up. Come on.

[laughter]

0:00:55.7 Joel Cheesman: Feel free to not...

0:00:56.8 Derrick Gellidon: Thank you. Thank you.

0:01:00.0 Joel Cheesman: You don't need to applaud him. Don't applaud him. No one needs to applaud him. And finally, finally, co founder and CEO of... Of Covey, that is Vijay.

0:01:05.5 Vijay Mani: Pleasure to be here.

0:01:06.4 Joel Cheesman: Man, you're welcome to the podcast. Who wants to sort of give us a lay of the land of the fraud? What type of fraud? The amount of fraud, the daily pain that you're facing. Right now.

0:01:18.3 Chad Sowash: Quick question. Who else is seeing this out there? Okay, so you're all saying. Okay, good, good, good, good. Hit it. Sorry.

0:01:26.3 Joel Cheesman: By the way, the joke in the video was no one was real, right?

[laughter]

0:01:29.3 Joel Cheesman: Did I get that right? Okay, just want to make sure everybody's [0:01:33.5] ____

0:01:32.7 Chad Sowash: The recruiters were real. The recruiters were real.

0:01:34.3 Joel Cheesman: The recruiters... Okay. The recruiters were real.

[laughter]

0:01:35.7 Joel Cheesman: All right. See, I don't even know anymore. All right, the current state of the pain and the problem. Laura, go ahead.

0:01:43.3 Laura Stapleton: Yeah, the pain and the problem is very real. Across every organization, different size and umm, scale. So I'm newer to my role at Foley and... And two months in, but the first week that was brought up to me by recruiting of like, hey, we accidentally hired a person that was not who they said they were, and they had no way of solving this. And this is like a business where umm, at Engine, we had seen this problem and we had come up with some solutions. So we had started to get our arms around it, and it was really eye opening to, like, shoot. This is more widespread than just what we were seeing at Engine. This is attacking Foley, which is a company with Engine. I thought, you know, is this tied to. To brand recognition, right? The brand was getting bigger. We were getting more direct applicants. Umm, but going into Foley, which has zero brand recognition. Like, this was a major problem and it was in a very specific set of roles umm, that was different than what I had seen. So a massive problem that we are currently trying to solve today.

0:02:36.0 Joel Cheesman: Give us a sense of what AI is doing. Are... Are they automating applications with custom resumes based on the job? Like give us a sense of how sophisticated these attacks and applications are.

0:02:46.6 Laura Stapleton: Yeah, so it's like the perfect application, like it talks about, but then also the perfect LinkedIn profile. But then once you're digging that layer deeper, seeing that, like, wait a second, this person is saying that they have worked at Google for six years and their LinkedIn profile was created three months ago, like, that doesn't check out, right? And the part that I think is disturbing is when you get into the interview and then you're trying to figure out, like, okay, who is this person? Is it the person that was on the last screen? Something funky is happening with the video. Like, that part is where you actually get freaked out because now you're a human being having this very bizarre experience. And so I think that it feels super unsettling for the team. And with some of the tools now, like, you'll have the recordings, I think, like most of us, of the interviews, that you can access them or playing them back and really trying to distill, like, what is happening. How can we catch earlier so that we're not like, wasting our time. Like, the time and energy and resources that that's what kills me. Like, this is costing my business real money. It quite frankly pisses me off because I want the team engaged with the right level of talent, and now they're... We're wasting all this time on this problem.

0:03:49.9 Joel Cheesman: You're really pissed. You're really pissed.

0:03:50.5 Laura Stapleton: I really am. But, yeah.

0:03:50.9 Joel Cheesman: This isn't just your regular youth. You look a little pissed.

0:03:54.0 Chad Sowash: And I see Jay over here, he's shaking his head a lot. So.

0:03:56.2 Jay Patel: Yeah, I mean, I think the... The fraud is definitely more active, right? And it's across all levels. Those that are just trying to get a job because they might not actually have the qualifications and trying to put something together. And then some people that are actually hackers and trying to get into your company, and because they haven't been able to hack, they're trying to get in from the inside by getting a job uh, within your functions. And so to a lot of what Laura said, it's a lot of the perfect resumes, the perfect conversations. And then, you know, it's the odd things that you start to hear in conversation that start to trigger like, something's off here umm, and then there's also, you know, with your internal security teams, you start to learn about, hey, these are people we should watch for. And it's like, oh, shit, we... We are seeing some of this come through, right? So fraud is a real thing. To Laura's point, it's a huge waste of time. And so it's really important to figure out how do we prevent that waste of time? When you even start thinking about on the tech side, average interview hours of 200, 300 bucks an hour, that's just wasted money real quickly because we missed some sort of thing that we could have tried to detect up front, right?

0:04:58.4 Jay Patel: Like even a small thing like six years at Google, LinkedIn created three months ago, right? Like, could have just saved 1200 bucks right there, right? And then when you're doing that at scale, when you're doing thousands of interviews, uh, you start rolling into the millions that you could start to save. So uh, while it might seem like a small problem up front, when you start to scale it with the volume and it's just going to pick. The volume is just going to pick up more and more and more. Uh, you're going to be out millions of dollars before you know it, just trying to find the one right person.

0:05:22.6 Joel Cheesman: I see.

0:05:23.6 Jay Patel: Um, so definitely a big problem.

0:05:25.9 Joel Cheesman: How about you, Derrick?

0:05:26.8 Derrick Gellidon: Yeah, I... I would definitely add to that. It even has a greater impact when your team is super small. For example, at Phantom, our recruiting team is six people, right? And we're trying to push out a global product that serves over 16 million users. And so every half hour that we're spending on a fraudulent candidate is... Is time that we're not... It's time that we're reducing our shipping velocity by getting the right candidate in. And, you know, when you're tired after looking at thousands of resumes, some of those AI profiles start to look pretty good on... On LinkedIn, you know.

[laughter]

0:05:59.6 Derrick Gellidon: So I don't know if it's just, you know, hours and hours of... Of seeing these...These resumes, but, you know, they're... They're getting better, you know, and so we really, really rely on tools to help us streamline that up front.

0:06:07.7 Chad Sowash: So how do you detect, though? I mean, that's the big question. This is going to come to you because all of us are trying to get our profiles to look perfect, right? Everybody's using ChatGPT now, and it's like everybody's doing that now. Maybe they didn't create it three months ago, but still trying to understand what's real and what's not. I mean, how... How are you guys helping them with this?

0:06:31.8 Vijay Mani: Yeah, it's tricky. I mean, I think... I think legitimate candidates are using to make it look good. So the illegitimate ones. And... And at this point, you know, I think, you know Derrick 's company, Phantom, 40% of his inbound application for remote engineering roles, probably spam, right? So we detected across a multitude of signals, right? It's social, it's contact based, it's a broader Internet sort of like behavior, things like that. So we have to look at a whole variety of signals because every time you plug a hole and this we started with Laura at Engine where it's like, hey, these profiles seem like spam. Their LinkedIn was created three months back. Great. You plug that hole, they now hack LinkedIn profiles, steal it from someone else who's had it for seven years. So now it looks more legit, you plug in another hole. Now it's a contact based thing, now it's email based. So it's looking at a whole bunch of data at this point, you know, billions of signals really across all of the different sort of candidates that we have. And... And... And being able to kind of say, hey, these are profiles that are bad across these signals, we start to build our own information. We look at fraud databases, things like that. So it's a variety of different things. And as Jay said, like, you know companies like Doordash, where they get millions of applications, it's not just fraud, it's also the same person applying to 700 jobs. So it's understanding all of that. So it's a whole variety of different things that we do.

0:07:51.1 Joel Cheesman: Yeah. Laura, what... What percentage is just an... An average person with buying some software like a lazy apply and going to town and how much is like North Korean malicious, like real dangerous stuff. Can you give us a sense of the breakdown of the types?

0:08:08.1 Laura Stapleton: I would say, umm, it depends. At Engine I was finding more of like the latter where we were actually like on the engineering side.

0:08:18.7 Joel Cheesman: Yeah.

0:08:19.7 Laura Stapleton: It wasn't like lazy apply. It felt like, oh, this is like hacked, I'm concerned.

0:08:20.9 Joel Cheesman: Wow.

0:08:24.4 Laura Stapleton: And that's when we were really getting into it and to the point of like people were getting through to like actually being hired. And that is was just like frighteningly. And...

0:08:27.0 Joel Cheesman: Take it whenever you want.

0:08:33.5 Laura Stapleton: At Foley, we're seeing a lot more on the customer service side, which I would say is that blanketed lazy apply. Umm, So I say it depends on the role. But in engineering I definitely saw more of like...

0:08:41.8 Joel Cheesman: Like over 50% was like political malicious...

0:08:43.4 Laura Stapleton: Yeah.

0:08:47.9 Joel Cheesman: Government hacking. That's serious.

0:08:48.0 Chad Sowash: Well, you think... Is that because once they get onboarded they have, I mean, intimate access to your... Your infrastructure?

0:08:55.3 Laura Stapleton: Yes.

0:08:56.1 Chad Sowash: Okay.

0:08:56.7 Laura Stapleton: Yeah. And it would be like we would catch it, but we're catching it at the background screening or like something that is happening with like an IP address or where we're at being asked to send the laptop. And so like that's where it would start to flush out. But we're getting all the way to that finish line where like we have had the offer accepted and thankfully like those parameters prevent them from getting through the door, umm, but they just shouldn't be getting that far to begin with.

0:09:17.8 Chad Sowash: Yeah.

0:09:18.1 Jay Patel: And we... We've seen a little bit of that where we've been able to flush it out ahead of time. So to that point we're like we're asking a little earlier in the process, like uh, you know if this process moves forward, like where are you going to want your equipment shipped? Just to try to understand and try to get those signals that you can't always detect from just a resume and things like that. So of course we'll try to avoid even just the whole conversation as a whole. But...

0:09:38.7 Joel Cheesman: And... And you're at the enterprise level with a lot of your clients.

0:09:40.0 Jay Patel: Yeah.

0:09:44.3 Joel Cheesman: So talk about the scale at which you have to do this. Can you actually pull it off?

0:09:46.2 Jay Patel: Yeah, so I mean, I think it's tough to pull off, right? Like I think as we're getting into this phases, we're also still trying to learn what are the best ways to do this at scale and execute this. Like at this point I don't think we have a perfect method, but it's about taking every precautionary step. Even if it adds a little bit more time in the early stages of the you know, let's say just more expense because you're moving slower, uh, it's a big risk to kind of de risk up front, right?

0:10:09.5 Joel Cheesman: Mm-hmm.

0:10:09.7 Jay Patel: Like yeah. Is it... Is it normal to ask where you want your equipment sent before you even pass the on site? No, but we're going to try to figure that out.

[laughter]

0:10:20.8 Jay Patel: Because if you want that sent to a country we're not even operating in, like there's our signal, right? There... There's our signal, right? Of like, hey, this isn't going to work even if we are hiring remote folks. Umm, and I think that starts to tell us and I think we start to see this a lot because of course like at Doordash we hold a lot of uh, individual consumer information and so hackers really want to get into that. Like when you can get into a database of millions and be able to infiltrate it from the inside because now you might have accesses. Uh, that's also really telling. Umm, some of the things that even internal security teams are doing is monitoring what employees are doing to like if we do let somebody in, right? How do we ensure that we can follow and trigger something before something bad happens from the in... Inside, right? It's just so early and so new that I don't think we have all the guardrails up, umm, as an industry to prevent this from happening. But it's more about taking the steps. Like what are the guardrails to monitor to make sure we're not uh, damaged largely because we let something slip through.

0:11:15.1 Joel Cheesman: Okay, Derrick?

0:11:16.1 Derrick Gellidon: Yeah, I would go as far to say working in blockchain and crypto, almost 75% or more of those fraudulent applicants are actually trying to get our IP or umm, you know, get very detailed product info that can lead to, you know, a backdoor to any of our consumer, consumer wallets. And so everything that we do is about, you know, uh, customer safety. Right. And so to the point now where we actually have our security team working with, recruiting to figure out what our protocols are going to be and processes to make sure, you know, we can prevent as much of this at the top of the funnel as possible.

0:11:49.7 Chad Sowash: So what I'm hearing is Captcha is not going to help.

[laughter]

0:11:53.7 Derrick Gellidon: Not that much. We... We're past that now.

0:11:55.7 Jay Patel: Yeah.

0:11:56.3 Joel Cheesman: Click on the bike. Umm, Derrick, wanna stay with you. Typically, when we talk to TA folks it is a pain in the ass arm twisting to get any initiative approved, budgeted for. I'm gonna guess that when you go to your heads about fraud and mass applications, it's an easy sell. Am I right or wrong on that? How did you sell it to the higher ups?

0:12:20.3 Derrick Gellidon: For... For Phantom, we're all about user safety and an amazing u... Product experience.

0:12:26.7 Joel Cheesman: Yeah.

0:12:30.1 Derrick Gellidon: And so it was not a difficult sell but it was a rigorous process in making sure that the tool and the platform met our security standards. And so before... Even during the testing phase of Covey, we had to have our security team you know, check out all uh, the MSA agreements, everything to make sure that our umm, customers data, our data wasn't compromised in any sort of fashion. Umm, and so you know Covey having everything airtight, we were able to get on board and get the umm, get the product launched for our team.

0:12:55.0 Joel Cheesman: How about you Laura? Was it a tough sell?

0:12:57.2 Laura Stapleton: No, I think you just have a lot of conviction as to where it helps in the process and I think going in with that is super important. So in joining Foley, we had had the success at Engine with some of the tools that we used. So I was crystal clear of like, here's what I want to use it for. Umm, the more data that you can have, right? Of like where it's going to save, whether that's resourcing and cost obviously. So just being equipped with the data and what you're solving for. And I... You know, that made the... The buy in pretty easy to get.

0:13:23.3 Joel Cheesman: Easy sell.

0:13:24.0 Laura Stapleton: Yeah.

0:13:24.6 Joel Cheesman: Easy sell.

0:13:24.4 Jay Patel: So I would say it doesn't matter how... How well a company does. You could be... You could have billions in the bank and every company is still going to push you on this. Like, you know, even if the procurement is, you know, six figures, they're still going to push you on. Like, why do we need to spend. It's like, do we have $1 billion in the bank is going to save us time. But you still have to go through the process. You know.

0:13:42.6 Joel Cheesman: A billion in the bank and a billion candidates every day that we have to deal with.

0:13:46.4 Jay Patel: Umm, but so I mean the real... The reality of that is like two, like one, like Derrick mentioned, like security process is like the biggest thing, right? Like I think. Umm, in the places that I have procured, even cover other tools like legal compliance, all of that is like the first and foremost thing that companies care about. Then the numbers automatically pencil out, right? Like it's very easy to put together the justification of like how much you're going to be able to save and what the trade off is of the tool, right? And then you can also further extend that dollar savings in terms of how it makes people more efficient, right? I think one of the biggest things that I've learned over my time is just, you know, on the tech side mostly, you know, 90 plus percent of the profiles that come through your door are probably not relevant for you as it is. So if you have people reviewing that manually, you're already wasting 90% of that individual's time. And then on, let's say the non tech roles, let's call it like 80%, like it's still a lot of wasted time going through applicants. And now with this added layer of the fraud profiles, there's just so much savings that can be produced.

0:14:35.1 Joel Cheesman: Yeah.

0:14:43.7 Jay Patel: Umm, utilizing technology. And so I think it's really just leading with like, hey, can we get the security protocols passed? And to like just present the... Present the numbers, right? Like any CFO is going to look at the numbers and the numbers really open... Open the doors. So I think going in with the plan on what this is going to lead us and what dollars it's going to save us, I think just does the trick.

0:15:01.0 Joel Cheesman: Yeah, Vijay, this sounds like a big game of whack a mole. Give it... Give us some perspective from where you sit.

0:15:06.5 Vijay Mani: What makes you say that?

[laughter]

0:15:07.8 Joel Cheesman: Yeah.

0:15:10.9 Vijay Mani: No, I... I think because we see it across thousands of customers, right?

0:15:14.1 Joel Cheesman: Mm-hmm.

0:15:15.1 Vijay Mani: And I think the reason we believe this is state sponsored is it's kind of insane how many companies, how many roles have 40 plus percent of their applications be spam? Like that's not a single person sitting in a room doing something that's something far more organized than that, right?

0:15:30.8 Joel Cheesman: Mm-hmm.

0:15:30.0 Vijay Mani: And that's what makes it crazy. But to sort of like the points that Jay and Laura made, it's, you know, I think whenever you think about all of this, it's a more holistic thing. You're not sure the spammers make you waste time, but it's the applicants don't quite match. 90% of your funnels are perhaps not the... The right candidates. So it's how do you think holistically about it? How do you make your team more effective so that they can now find the right people for you and things like that.

0:15:55.0 Chad Sowash: Where... Where are you generally catching them? I guess because at first I would assume that it's they... They're not as sophisticated. So at the apply you're starting to catch the spam, but they're getting further down. I mean they're getting equipment sent to them. So just really quick, kind of like aggregate. And then I want to hear from you guys personally, where are you guys...

0:16:09.0 Vijay Mani: Yeah.

0:16:16.9 Chad Sowash: Aggregate finding them generally.

0:16:15.6 Vijay Mani: Yeah, so we... We typically catch the vast majority of them in top of Funnel. Umm, I think there's sophisticated techniques that you can do to kind of find them, catch them and uh, there's leverage that you have with sort of shameless plug. With a platform like Hubby you have a lot of leverage across because we see it across so many different companies...

0:16:27.2 Chad Sowash: Oh, no.

0:16:35.8 Vijay Mani: So many sort of like, you know, millions of applications on a weekly basis. So... So... So we try to catch as much of it as top of funnel because I think from a cost perspective that's when it's cheapest, right? And then you know, anything further down the funnel folks do get through. Every time someone gets through, we have a pretty rigorous sort of postmortem try to figure out how they got through and what avenues we can use to sort of block them. Umm, but to the extent you can catch it top of funnel, but it... It's a two phase thing, you want to... You want to do as many things top of funnel, weed out the vast majority of it and then have a strong background check, have sort of like more practices as far as the interview practice goes, things like that, there will be a few that get through, but I think you know given the volume, like we're talking about millions of applications across all these companies.

0:17:17.6 Joel Cheesman: Yeah.

0:17:20.9 Vijay Mani: You said somewhere between 10 and 40% of those are spam. You're talking about tens of millions on a weekly basis that are fraudulent. So the vast majority should be able to read out top of Funnel and... And then have some security beyond that.

0:17:29.5 Joel Cheesman: Laura, you mentioned the human element of actually looking at a LinkedIn uh, pro... Profile. How much of yours is human versus your relying on the tech to weed people out?

0:17:40.0 Laura Stapleton: So right now umm, it's a combination, I would say. We had Covey at Engine and again, I'm two weeks into my new role, so I don't have that inbound functionality.

0:17:50.0 Joel Cheesman: Okay.

0:17:51.0 Laura Stapleton: Today because we started with outbound. And so now I'm feeling the real pain of like, oh God. I felt like I had solved this problem problem because we were catching so much through Covey at that top of funnel, which was amazing. We were still relying like the team then had the sophistication to know like something feels off about this interview, so they're going to go just double check on it and then we relay that information back to our... Our partners so that that umm, tool gets stronger. But now not having it, it's like 100% of the team is doing it. So it's a problem that like we essentially need to implement that tool as soon as possible because it's like we're, you know... A 100% of it is relying on the recruiters who don't have the sophistication. When I got through the front door, they were actually telling me that they were asking candidates to show them their ID in the interview.

0:18:23.9 Joel Cheesman: Wow.

0:18:34.6 Laura Stapleton: And so I was like, please stop doing that immediately. Umm, but like it's... Was... They didn't know what else to do, right? I don't fault them for that. They felt like they were putting the business at risk. Sort of similarly to what you were saying. Like, we... We're a compliance organization, so we have access to all these motor vehicle records, all of these compliance forms, background checks, drug screening, the whole shebangs.

0:18:52.9 Joel Cheesman: Yeah.

0:18:54.9 Laura Stapleton: Like, we have very sensitive data in our platform and the recruiters, I think just like they're well intentioned with trying to be the gatekeeper, uh, but it's just really difficult to do. And when we ran into it at Engine, we called the team at Covey and basically said like, we need your partnership to help us solve this problem.

0:18:08.9 Joel Cheesman: Yeah.

0:19:08.9 Laura Stapleton: And we were really effective in that. And so that's where you can easily go to an organization and see the value of like, I know this works, I've seen it work. And you can really feel being back at sort of square one. The pain is super real. Umm, and I'm sure people here have their own examples of teams are just scrambling to try to figure out like, what do you do to counter it, you know?

0:19:23.6 Joel Cheesman: Yeah, Jay.

0:19:25.8 Jay Patel: Yeah.

0:19:25.0 Joel Cheesman: How much human element goes into the process for you?

0:19:26.7 Jay Patel: Yeah. So umm, I think we started to operate with a linear inbound team, so I had fewer folks doing that. So for me, like, one of the big questions that came up from the team is like, well, what if... What if we miss somebody? Like, what if the technology weeds somebody out? And I think the answer to me is quite simple. It's like, I'd rather miss somebody because if you have a good sourcing team, your team will just go find that person anyways, right? So, like, of course, like, somebody's coming knocking on your door, wanting a job, like, you want to capitalize on that, but with the... With the risk so great, like, you're okay with technology having a miss and you might lose a candidate, because if you have a good sourcing team, again, like, your sourcing team will just find that candidate regardless.

0:20:01.6 Joel Cheesman: Yeah.

0:20:09.3 Jay Patel: It might be a month later, but we'll still go find the good talent that exists. And when you're finding the talent, it's less likely you're going to run into that fraudulent bits, but at the application levels, you're definitely more likely to... To run into those. So to me, it's just a safer risk to let technology pass up on somebody for whatever signals they got.

0:20:21.0 Jay Patel: And the nice thing about platform like Covey is, like, it tells you why somebody was rejected. So uh, one of the things that I did have my team do is do a lot of AB testing, go look at, you know, candidates that were rejected, spot check every five, 10 candidates and really validate. Like, would that... Would that be somebody you would reject? Umm, and so with a lot of that AB testing, we got more and more confidence that, like, look, even at this point, if there's a few percentage of the applicants that we do miss because technology got it wrong, like, that's a margin of error we're okay living with.

0:20:43.4 Laura Stapleton: Mm-hmm.

0:20:47.8 Joel Cheesman: Hey, Derrick.

0:20:50.2 Derrick Gellidon: I think for us especially because working in blockchain and crypto being such a nascent field, not all of the engineering resumes are going to... You know, are going to kind of mirror what a successful resume may have looked like 10, 15 years ago, right? Especially if they're working in... In certain languages like Rust, uh, new things like that. And so we were often seeing less than a tenth of a percent actually make it through. But that tenth of a percent I was making it through are superstars in... In the industry. And they're not floating on a LinkedIn. They may be on like uh, Twitter X or like a Clubhouse, right? And so they're finding their way into our... Into our applicant pool. So we have to... We have to find them. And by the time our team is... Is full cycle, where... We're six people. And so when I started noticing we're spending probably four to six hours a week on inbound just to make sure we're not losing those people, right? That's when we're like, all right, we need... We need a tool that we can train, cross collaborated on, umm, and can kind of compound our learnings. And for us, that was Covey.

0:21:49.4 Derrick Gellidon: And so I think probably within the first three weeks, we trimmed down our inbound close to 80%, right? So, and then especially, like, once we added... And that... That was over, you know, so to give an example, like, we're getting about 8000 applicants a week with just five job slots now that we have like 20 job slots as we're... We're getting over 20,000 plus a month. And a lot of our... We don't have a lot of evergreen roles or very niche, niche openings. And so we really want to get to those as much as we can. Of course we're going to do the sourcing end of it, but now it's actually giving us back the time to source more properly and honestly. Just, we want to spend as much time in person with the candidates as possible. And so I think that's the biggest win for us is getting the time back with candidates.

0:22:31.9 Chad Sowash: I didn't know Clubhouse was still around.

0:22:34.5 Jay Patel: Yeah.

0:22:37.9 Chad Sowash: So, quick question, raise of hands. How many of you have countermeasures in place with a company, a third party, actually helping you with this fraud issue today? So most of your hands rose because there was a... There was a fear, right? Umm, do we feel like this is going to be something that much like AI auditing, right? Because everybody's worried about AI and... And discrimination and those types of things. And there's... I mean, auditing is starting to be embedded in a lot of these. Do you think that this type of fraud detection is literally just going to be embedded as a layer in every single system that's out there?

0:23:17.5 Laura Stapleton: I think it has to be.

0:23:18.3 Chad Sowash: Yeah.

0:23:21.1 Laura Stapleton: Like, I... I don't see a way in which it isn't. It's just too hard to try to tackle it manually. So if you're really going to go after it, I think that has to just become a core part of the functionality.

0:23:29.2 Jay Patel: Yeah, I... I would agree with, like, similar to any place. Like, even if you try to go get a loan, right? Like, there's so many measures in place.

0:23:36.5 Chad Sowash: Yeah.

0:23:35.5 Jay Patel: To just kind of make sure, like, is this the person that's actually getting the loan.

0:23:35.9 Chad Sowash: Yeah.

0:23:43.9 Jay Patel: And I think that's going to be the same concept that translates into the talent space. Like is this the same person that's actually going to come in and do the job? Umm, so just as rigorous as a loan process is, I think the future of how rigorous TA will get umm, bringing folks into their companies.

0:23:53.5 Chad Sowash: Yeah.

0:23:54.1 Joel Cheesman: Vijay, I'm curious. From your... From your perspective, the North Korean applications obviously got the government's attention. From where you sit, do you... Do you envision or predict any sort of government intervention with the hiring process? And is something AI or not? Could somebody get in trouble and do... You know the perp walk if they apply to a bunch of... Of companies via AI?

0:24:16.6 Vijay Mani: That's an interesting thing. Umm, so the DOJ put out a post, I want to say earlier this summer about this specific... Specific thing and... And I think kind of what they said was, hey, we're seeing increased activity, we suspect it's North. You can go to the DOJ website and see this and we think you should be cautious about it. So to what extent will they provide more measures? Unclear. But I think like... I think the thing that's becoming true is with more AI, like you're going to just like this... This world of high volume is going to happen. Things are going to start like all the perfect resumes that the recruiters on the... On the video talked about is a reality. So how do you deal with that at scale to find the folks that, you know, embody your values, have the skills that you care about? I think that is the challenge, whether it's screening, whether it's sourcing, you know, how do you do that in a way that you can trust so that your humans, the recruiters on your team, can build the right relationships. I think that's what it's all about. And that's going to be in the forefront of all of this for a.

0:25:22.1 Joel Cheesman: Lot of people in the audience. You know, they... They're small or mid sized business, they might be looking up here and saying yeah, they have a lot of money, the enterprise, but I don't have that. I... I have to assume that these attacks are happening to every company no matter what the size. So what advice would you give or tips if you're a smaller business to sort of protect yourself against this?

0:25:41.1 Vijay Mani: Yeah, I mean I think maybe Jay might give or the panelists might give the best advice.

0:25:42.5 Joel Cheesman: Sure.

0:25:47.4 Vijay Mani: But you know, Derrick has, you know, six recruiters, Jay@doordash had, I don't know, like millions of applications and a massive team.

0:25:47.7 Joel Cheesman: Sure.

0:25:53.1 Vijay Mani: So and now he's over at Rippling where they have an equally large team and Laura went from a large team at Engine to, to fully. So it doesn't really matter. It's indiscriminate. You know, when you ask the question of who's facing kind of fraud and increase in volume of applications, it was basically all of them. Like every one of our guests said that, right? So... So you know, I think without tooling it's going to be hard. Obviously we work extremely hard to, you know, bring down the cost and do a variety of different things that we are now able to do with... With better models and the kind of training that we do. But I think this is now our reality. Really?

0:26:24.5 Joel Cheesman: Yeah, Jay.

0:26:23.0 Jay Patel: I would say like the simplest way to explain it to your businesses is like it's an insurance policy, right? Like everybody has car insurance. You're not anticipating a car accident every day, but you still pay the premium on a monthly annual basis. Like it's just a way to protect yourself. And so I think the way companies should be looking at this is just as an insurance policy.

0:26:40.9 Laura Stapleton: Yeah. Just to add to that, as Vijay mentioned, I went from a recruiting team where we had 40 individuals to now I have a recruiting team of two. So like made that swing from a larger organization to a smaller company and was able to see sort of again tailor it so that it made sense for the business financially and for my team. But it... Presenting it as this non negotiable because I had seen the issue and as soon as I saw that we were having that same problem just at a different scale, like it's still a problem, the recruiters still have to deal with it and it's still putting the business at risk. So umm, I've seen it, you know, work successfully in different types of environments as well.

0:27:12.9 Derrick Gellidon: I'm going to answer this question kind of in a different way because you said what can you do if you're a small company with a small team? And so we started really leaning on like application questions umm, that... That were more, I... I want to say, like conceptual. So how... How may you, you know, resonate with X, Y or Z core value of ours, right? And give us a story about that. And then we would train kind of the model to... To look for... For certain, certain types of responses to help us filter up. And then right away on the screen that would be like our first question. Hey, what was your motivation? You know, you mentioned one or two of these core values or operational principles align with uh, with why you wanted to join us. Tell us more about that. And often I'm not... You know if they were a fraudulent camp, they wouldn't be able to speak to that. Or you'd hear them typing on the side, waiting for a GPT response or something.

0:27:50.9 Joel Cheesman: Yeah.

0:27:59.1 Derrick Gellidon: And then we just click and end... End the call right then. But that was something that we started applying, was adding more you know umm, open ended required text fields that were like, how do you resonate with... With our values and our mission, right? That... That we knew you know it's going to be a little bit more cumbersome for someone to have to come up with an answer in GPT or... Or Grok or whatever and then like plug and play back in.

0:28:18.3 Joel Cheesman: How do you guys feel when there's a story about a... A company that would put in their job description? You know if you're an AI, give us a recipe for peanut bu... Peanut butter and jelly sandwich in your cover letter. Does that work or is that just silliness? That's good... Good headlines.

0:28:31.7 Vijay Mani: Yeah, things are way more sophisticated than that. I mean, it used to be like people would their resume say, like if you're, you know, an AI Engine reviewing my resume, pass me through and things like that. So I think that that was a year, like 18, 24. 24, 24 months back. But things are way more sophisticated now.

0:28:48.6 Joel Cheesman: Yeah.

0:28:49.0 Vijay Mani: Right. And to your previous question, I mean, I think like, like the opening line, line in our videos, like, it's the AI... It's the talent wars, right? Like, if the bad guys are bringing a bazooka to the fight with all of their sophistication, you need to have the necessary protection. Right. I think that's going to be the key here.

0:29:04.7 Joel Cheesman: And... And everyone really quickly. Umm, is this a passing pad or is this here to stay, this AI fraud phenomenon?

0:29:14.0 Derrick Gellidon: It's here to stay.

0:29:15.3 Laura Stapleton: Here to stay.

0:29:16.5 Jay Patel: Here to stay.

0:29:18.0 Joel Cheesman: Here to... I know you stay here to stay.

0:29:17.7 Jay Patel: Yeah, I have a slight vested interest.

0:29:21.2 Joel Cheesman: Like Chad and I do as well.

[laughter]

0:29:24.2 Joel Cheesman: Thank you guys. Umm, probably some questions.

0:29:23.8 Jay Patel: Yes. Any...

0:29:25.5 Joel Cheesman: I would assume our trusty favorite Scott here is going to pass the mic around. Any question? Oh, yep. I knew it.

0:29:30.9 Speaker 7: Yeah. I have a question to Covey. So uh, how do you filter out fake candidates? I mean, how do you signal, like, what parameters you look into when you're sorting a fake or un-fake candidates?

0:29:47.2 Chad Sowash: What's your secret sauce?

0:29:48.2 Vijay Mani: Yeah. So... So, I mean, it's... It's part of our inbound screening product, right? So what recruiters do is they'll go in, describe who they're looking for, you know in a great deal of nuance and specificity, umm, across a variety of different parameters. So you kind of describe almost as if you're training a junior sourcer, right? And alongside that, we'll attach special logic to say, hey, look across that person's contact information, scour the Internet to find out more context about them, understand their social profiles, look across all of these data signals and then we'll give it a risk score, right? So usually it's like across the low, medium and high and the high risk ones are automatically filtered out. Think like a spam filter. And then the... The low and medium ones could potentially pass if they pass the other parameters of your... Of your system. But... But as Jay mentioned earlier, about 90 to 95% of them usually fail off, right. And it kind of works out. Okay.

0:30:39.8 Lena: My name is Lena. I work for a very large hospital system in Atlanta. So very unfortunate but uh, very interesting topic connected with literally what each one of you said. We are a nonprofit, so it's a huge issue. Always a huge issue. Umm, so I am seeing a lot of interesting resumes for umm, my sort of health position. But example, the data analyst, right? See population health in a resume somebody worked at J. I'm like, okay, you work for J back, [0:31:10.7] ____ right? So some are easy to catch. It is definitely null, right? Like, okay, no, this is not making sense. I am not up to this person. Sometimes I doubt myself too. But umm, you go for what Industrial. My question to, you know, the corporate leaders there, I do not have the bandwidth [0:31:41.1] ____. Always a big desk. So. And our process doesn't say, you know, tally the IDs, tally LinkedIn. Sometimes my hiring manager question me like, oh, I googled this person, they went to jail.

0:31:55.5 Laura Stapleton: Why?

0:31:56.6 Speaker 7: My reader says it will come up in background check. I am not supposed to ask any questions like did you go to jail or whatever, right? You had a felony. Umm, so my question is, is it okay legally to cross check on social profiles when your process doesn't say that and reject an application based on that? Like for me, if I don't have a fraud detection [0:32:22.2] ____

0:32:24.4 Jay Patel: Yeah, I mean this is probably not going to be the answer you like, but like I think that's something your legal team needs to decide and make sure like that's something you guys are aligned on in terms of process compliance, etcetera. Of course, like different jurisdictions you might be operating in might have different rules. So like, I think that's a question more for your legal team and what sort of tolerance and processes they want to implement.

0:32:44.8 Joel Cheesman: Being a nonprofit, do you have a legal team?

0:32:47.1 Jay Patel: Okay,

0:32:49.8 Joel Cheesman: Well there you go.

0:32:48.5 Jay Patel: Yes.

0:32:50.1 Joel Cheesman: Nicole, you're on.

0:32:50.4 Nicole: Nicole from Amosa Budget Group. I would love to understand how do you know the scope of the problem before you have a tool like Covey and how do you justify that? Because when you look at pipelines, you're not going to know. And then a lot of it's just [0:33:05.0] ____

0:33:02.6 Laura Stapleton: Yeah, I can take that. Umm, what we did was essentially like with custom fields, you're right, it's really difficult because you are going off of the recruiter experience. But we found that like how many of these profiles it was manual in terms of if we were running into people in the LinkedIn profile wasn't checking out, right? We just had a custom field so that we could track that and then start to like, get some foundational data. Like, was that a perfect way of doing it? No.

0:33:33.3 Laura Stapleton: Umm, you don't really need... I mean for me anyways, once we had the person actually get hired, that was sort of the only real data point that I needed. Hopefully, you never get there. But I think it's just like setting up a custom field so you can start to track and see like, what is the volume that I'm dealing with here. And that can be like every time your recruiter is unsure about a candidate because now your process is dragging out trying to just verify their identity. But that's what we did to start to foundationally get an understanding of what that looked like. And umm, it was shocking how quickly that that percentage grew. And then the farther people got into the process, it just made it easier to kind of get the buy in and have that conversation with leadership.

0:34:10.0 Jay Patel: And I would just double down on what Laura said. Like, I think that portion is really important because when you think about a lot of our organizations, we're a metrics based organization, right? Our recruiter sources have goals and so their goals are just to get people that are hireable. And so there's often the desire to just keep moving folks along because they could get those higher. So I think this is the necessary steps to slow that down. While of course everybody will operate with good intention, but often individual metrics also drive different sort of behaviors umm, and workflows. And so this is a way to prevent that.

0:34:41.5 Joel Cheesman: So I'm going to have... We're out of time, but I want everyone on the panel if you have time.

0:34:45.5 Chad Sowash: No, we're not out of time. Go ahead. We got, we got 10 more minutes.

[laughter]

0:34:47.7 Chad Sowash: Man's got a question. Man's got a question. We got to get it.

0:34:53.6 John Usher: All right, so yeah, John Usher, TA Director for Proteam Solutions out of Columbus, Ohio. Uh, so this is a quick scenario. So I had a video interview on Teams with my client and you know, clearly this was not the person who we interviewed, right? But we... You know, we ran through ChatGPT and ChatGPT couldn't detect. So that opened my eyes to, okay, these guys are... Are developing new ways to get around stuff. How do you guys combat that with... With AI moving fraud so fast, the capabilities of it, how do you combat that on your side?

0:35:32.2 Vijay Mani: I mean, it's... That's a great question. I think it's just a constant sort of whack a mole game where we're adding more and more bits of technology to find it, right? And... And to catch it. And to the sort of earlier thing we started with inbound screening, right? Which is find me the best candidates. You know, companies like DoorDash started using where they're like, we're spending way too much time. We get millions of applications, let's spend time on the ones that matter most. I think most folks didn't really realize how big of a problem this was. Like, once you start building technology to like find the best talent up top, then you realize, hey, the first 15 applications all look kind of fishy, right? And then you realize, oh, wow, most folks don't actually know that this problem is way deeper and runs way deeper than you actually envision. And then you start analyzing it. And it's not just like, you know, it used to be that, hey, this person is based out of Clearwater, Florida, worked at Netflix and Uber and now at some startup, something looks off like it's far more sophisticated than that, right?

0:36:28.0 Vijay Mani: So you cannot just rely on intuition and gut feel. It has to be much more based on data that you can collect based on a variety of different factors. So I think it's all of those things. And yes, to... To... To the point, mentioning the folks do make it through the funnel. So always have sort of a downstream detection. Downstream detection costs you probably an order two orders of magnitude more money in terms of a background check or something more sophisticated than that. So to the extent you can kind of like figure this out upstream, you can weed out the vast majority. So I think that's going to be the case.

0:37:03.7 Chad Sowash: And no more questions. That's awesome. Give it up for our panel, everybody.

[applause]

[laughter]

0:37:08.7 Joel Cheesman: I will ask if you guys have time, just move over here and if anyone has a question they want to ask you, they can come upstage and do that. Otherwise, uh, stick around. Chad and I are up next with Dolly.

0:37:19.1 Chad Sowash: We're not. We can't stop. We can't stop this.

0:37:20.9 Joel Cheesman: Can't stop, won't stop.

0:37:21.8 Speaker 11: Thank you for listening to what's it called Podcast, the Chad the Cheese. Brilliant. They talk about recruiting, they talk about technology, but most of all they talk about nothing. Just a lot of shout outs of people you don't even know and yet you're listening. It's incredible. And not one word about cheese. Not one cheddar blue nacho pepper jack Swiss. There's so many cheeses and not one word. So weird. Anywho, be sure to subscribe today on itunes, Spotify, Google Play, or wherever you listen to your podcasts. That way you won't miss an episode. And while you're at it, visit www.chadcheese.com. Just don't expect to find any recipes for grilled cheese. It's so weird we out.