Your credentialing specialists can spend up to an hour on every placement checking a candidate’s documents against a requirements list. Even the most experienced ones still get it wrong sometimes.
Eric Ly, Co-Founder and CEO of KarmaCheck, joins David Folwell to explain how AI is taking that work off their desks in healthcare staffing. He also covers why AI is making hiring fraud harder to stop, where people still need to sign off, and what to ask before you buy back-office AI.
[0:01:13] DF: Hello, everyone. Thank you for joining us for another episode of The Staffing Show. Today, I’m extremely excited to be joined by Eric Ly, who’s the co-founder and CEO of KarmaCheck. Eric, great to have you back on the show. Excited for the conversation today.
[0:01:27] EL: Yeah, same here, David. It’s been a while since we last spoke on this topic. So, great to see you.
[0:01:32] DF: Yeah, it’s going to be fun to dig in. A lot has changed. I think looking at the records, and last time we talked was 2023 on the podcast, and ChatGPT was encroaching, and there are things that were happening. And we’re all trying to figure out what was this world going to look like. And we’re going to dig into that today, as well as what’s going on in AI use cases within the back office. Before we do that, Eric, give a little background on kind of who you are, if you want to tell a quick story on how you got into staffing, and then we’ll jump into some questions.
[0:02:02] EL: That sounds great. So I am a tech founder by background. I’ve started a number of companies.
[0:02:10] DF: May have heard of it.
[0:02:12] EL: Thank you. Thank you. Yeah. The most well known of it, of which is LinkedIn, where I was one of the co-founders. That really led to the current company, KarmaCheck, because we were just very interested in instilling more trust with respect to identities. And so KarmaCheck has been working in the background check space ever since the beginning. And specifically, we’ve been working in the healthcare space for most of that time.
[0:02:39] DF: I actually haven’t heard the kind of the origin story for KarmaCheck and how that tied to LinkedIn. What were some of the elements that made it so that you realize like, all right, well, this is a thing that needs to exist?
[0:02:49] EL: Yeah. So several years ago when KarmaCheck got started, a bunch of us, I’m one of the co-founders of KarmaCheck, but we were just very interested in trust and bringing greater trust online. That was a time where issues around fraud was definitely coming online. You could definitely see it as a trend. Maybe we were early in recognizing those trends, but having been part of a company like LinkedIn that got professional identities online, we were wondering to ourselves and asking, “How much are those profiles and identities actually authentic and accurate?”
And to my surprise, I’m obviously biased. But to my surprise, they were much higher percentage of identities and profiles that have inaccuracies in them intentionally, right? Intentions to deceive. And we just really thought that when you have authenticity, when you have trust, it leads to a lot of great things in terms of opportunities for people, new jobs, transactions, business relationships. And so we knew that trust was really important. And so we wanted to add our part to it.
And so KarmaCheck was designed to really provide kind of the verification piece of that or in other words background check piece of that, and we wanted to attack the hiring space first and create more trust around hiring. And at the time, background checks, we saw that there were just lots of opportunities to improve how background checks were done. Really taking what we knew in terms of automation and technology, those kinds of tools to make background checks go faster, better, and cheaper.
[0:04:30] DF: And so you saw it really early. And I know background checks have been around for quite a while. But did you know that this was going to happen? I mean, since we’ve talked last with ChatGPT coming online, AI coming online, and now the quantity of fake resumes, fake applications that people are seeing in the staffing industry, I think is at an all-time high. Did you see that when you were starting KarmaCheck?
[0:04:52] EL: Well, we certainly knew that. And we knew that when we first started, AI wasn’t really as much of a thing. It was a few years later that AI came onto the scene, kind of burst onto the scene. And it was certainly at that moment that we realized, “Oh, this is not going to be a problem that lessens now. It’s going to be a problem that increases because the tools that are available to both the bad actors as well as the good guys who are trying to protect the integrity of these processes, it’s just going to become more of a problem, basically. And so it’s very much using these tools between the good guys and the bad guys to battle the increasing fraud that I think is much of an issue today in the staffing industry.
[0:05:40] DF: Yeah, I’ve heard anywhere from 20% to 40% of applications are coming from AI now. I’ve even heard that one out of four will be through AI in the next couple of years. And I think that that obviously increases the likelihood of fraud when you have tools out there applying for you. For those of you that are listening, there’s ways to replicate your voice with tools like ElevenLabs. And there’s things out there that are happening that are exciting and also simultaneously scary.
One of the things that I think is interesting about what you guys are doing is you’re digging into the background space, but you guys are also using AI. And while most of the people that I talk to with their AI use cases are really focused on front office sourcing, qualifying, interviewing, and trying to find new talent. You guys are kind of finding use cases for that in the back office. What do you think is especially interesting or exciting about the different use cases when it comes to the back office and AI?
[0:06:35] EL: Yeah. Certainly, I think one of the things that has been about AI is that it is pervasive. It’s going to be pervasive, even more pervasive than it has been in the past. It’s going to kind of infiltrate all aspects of technology and the way that we do things, the way that business and daily lives get run. And so I think that was one of the surprises, going back to one of your earlier questions about what did we kind of guess right or what did we not guess that is happening now in the world?
And so it’s clear that AI is definitely a very useful technology and tool that can improve lots of things. And so there’s definitely a lot of sexy applications out there. But being in background checks, we’ve always kind of focused on the back office. And when these fabulous tools became available, we asked ourselves, “What can you do with these tools in our world?”
And we definitely saw that there were just a ton of things on the back office that we could apply AI technology to make our customers’ lives, the employers who are running background checks, a whole lot better, like a whole new level of efficiency and productivity that wasn’t possible before. And I’ll just say that in terms of administrative costs and efforts, and we do a lot of work in healthcare, there’s really just a lot of opportunity to apply this to bring about efficiencies over the next few years.
[0:08:04] DF: To dig a little bit into KarmaCheck specifically, what are some of the things that you guys are doing with your customers or some of the use cases where you’re finding it impactful for them?
[0:08:13] EL: Yeah. One of the things that we have found, we do a lot of our work in healthcare and healthcare staffing, and that’s really our main focus in this realm of healthcare staffing, the process of doing background checks, or as our customers would like to call it, credentialing. It’s really a complicated process on so many different levels. It’s not like your typical kind of commercial staffing background check where maybe there’s a few things in credentialing a clinician, be it a nurse or a doctor, there’s often tens of items that need to be checked in this process.
And so it’s really complicated and it really changes a whole lot. And so it makes the work of a staffing company that much more complicated. Credentialing is something that is one of the main concerns when you have a staffing agency or a staffing platform. And so we’ve seen things like handling of documents. When you’re thinking about getting someone placed into a role, there’s a lot of documents that change hands coming from the candidate themselves that need to be submitted.
There are also documents that come from the verification bodies, the verification sources that they submit. That needs to be handled as well. And so we find that a lot of the time that it takes to get someone through a process, that credentialing team or that credentialing specialist, a lot of time they’re handling documents. And that is something that with the advent of AI can really be automated and takes that burden off of people having to do it.
They would, I think in many cases, rather not do it because the interaction with the candidate, as lots of people have talked about, is really where the valuable part happens between a specialist with the candidate. And so it’s things like that that AI can take over now that really leads to some really significant efficiencies. And we think it’s going to be the way that things are done in the next five years.
[0:10:17] DF: Absolutely. And when it comes to some of those background processes, which ones are you finding you’re able to have the biggest impact on? Is it the requirements, documents, follow-ups? Are you helping on the renewal side or is it just on the kind of background check specifically?
[0:10:32] EL: So one of the really interesting things about credentialing, and we’ve become nerds around this because it’s just so interesting in a way, is there’s just multiple levels. And one of the unique things that staffing firms in healthcare have to deal with is this sort of so-called requirements, handling requirements that come from hospitals, they may come from MSPs, and then there are regulatory bodies like Joint Commission that have their own requirements. And the agency themselves has their own standards that they need to admit.
So anytime there’s a placement, there’s this thing called a requirements checklist that needs to be gone through. And for each placement, what does the candidate – what have they fulfilled? What have they not fulfilled? That’s different in every case for literally down to the place where the clinician is going to be working. It’s going to be different. And so that’s been a pain point that credentialing teams spend a lot of time on. They could spend upwards of half an hour to an hour for every placement, just kind of understanding what it is comparing what has been submitted from the candidate versus what’s not. And sometimes they even get it wrong, even for the most trained person.
And so hopefully people get a sense that that’s a very complicated, error-prone process. But it is a perfect opportunity for AI and LLMs to understand those requirements which often comes in as PDF documents or spreadsheets that are basically not very structured. Yeah, pictures, PDFs. But it’s great for AI to understand that and to interpret that and to structure that information so that it becomes very much of a science and very deterministic so that you take the failure or the misinterpretation out of it, and it can become a very predictable and fast process. And so we see opportunities like that to apply AI in some very meaningful ways.
So we’ve seen, in terms of customers who are using the stuff that we have today, do things twice as fast as they have before without the AI. And we think that as we develop this technology, the improvements are going to be even more amazing than that. Imagine getting 80% of the efficiencies from using AI than before. And so what this means to the team is that they can now be much more productive. They can handle more placements. And it could be the same team that’s handling it all. So they’re overseeing the process as opposed to having to laboriously and manually do these processes. And I think that’s where the big win is going to come from.
[0:13:22] DF: This is a bit in the weeds on this question, but I’m curious in terms of the talent experience and what their expectations are. And if you’re measuring the impact that this approach is having compared to a traditional approach.
[0:13:35] EL: Thanks for pointing that out. AI is really going to help save on the administrative work. And that applies not only to the back office with the compliance teams and all, but it also applies to the candidate. So you can imagine a candidate on the other side of this process having to submit these documents, getting them collected. Sometimes the documents that they submit, while they’re well-meaning and well-intentioned, may be the wrong documents or may be expired.
And so putting in AI into this process to help candidates sort of through this process, which is arduous for themselves as well, is going to be very helpful as well. So it’s going to smooth over the credentialing side of the process for the candidate very much as well. So there’s just going to be wins from both sides.
[0:14:24] DF: That’s great. That’s one of those areas where travel nurse might want to talk to a recruiter about their next assignment, but if they can move through this part fast, I feel like this is such a good use case for AI, where it’s like, “Just give me the answers and help me get done what I need to get done as quickly as possible.” Do you guys manage the process with the candidate? Is it through text? Is it web-based? Or how is that?
[0:14:43] EL: Well, there’s really an opportunity to tie all of these different mediums together, right? So we’ve seen some really exciting uses of technology for interactions with candidates, right? Candidates come in through email, they come in through text, they’re on a call with recruiters or specialists these days to handle those documents. And so there is the opportunity for this AI technology to work across all those mediums, whatever is best for the candidate, actually. According to the times during the day when they’re available. When you have a nurse, maybe they have a very busy day. They definitely do. And so it’s often during the late hours or the wee hours where they have a few minutes or they’re driving to someplace. And so if they can take care of this stuff conveniently with very low time, that’s going to be a win. And for the employer, for the staffing agency, that’s going to be a big win, right? Because it cuts down on the friction to interact with that clinician that much more.
[0:15:48] DF: Yeah, absolutely. Switching gears a little bit and talking about kind of your team and how you guys are using AI, where are some of the areas where you’ve seen AI work for your team and what are some of the things where you’ve seen maybe it’s not having the impact that you expected?
[0:16:02] EL: Yeah, we’re certainly offering AI products. But even more than that, the way that we started was we actually implemented AI behind the scenes, without our customers even knowing, that automate certain manual processes that we’ve had to deal with in order to deliver our background check product to our customers. And we started to put AI into those processes to automate them and honestly get better results.
And so we’ve been quiet about it, but we’ve been introducing more and more AI behind the scenes to make phone calls, to verify documents, some of the things that we talked about, that we can offer as products to our customers. But we’ve tried those first and it’s just led to a much better consistency, fewer complaints from our own customers about, “Hey, why didn’t you handle those things? There are mistakes in these documents that we get.” We handled those first even before customers see those documents. And that’s done more and more with AI.
And so the opportunity to catch mistakes earlier, to have a faster process that gets results back to our users much faster is definitely part of the win. So we’ve definitely seen dramatic wins on our side, on the inside, with more opportunity to go as we develop more of these technologies.
So it’s going to kind of be everywhere. So that’s kind of the exciting piece. And I think that the challenge is going to be just making sure that AI works well with people and that whole kind of interaction is designed correctly between people and the technology to have just the right results and to engender the right amount of trust between the people and the AI.
[0:17:52] DF: It’s funny, I think you and I talked briefly about this, but I remember the first time I heard the idea of the human-in-the-loop and I was just laughing about this concept. What does that mean and how are we going to engage with AI? My view on that has changed pretty drastically over the last year as we start to get into it. What’s your take on where a person should sign off today and how do you kind of see that evolving?
[0:18:13] EL: Yeah, I think it’s still early days, which is to say that I think that people today, as they work with these AI systems, whether it’s ours or others, they want to have more control. They want to oversee what the AI is doing or what it’s recommending, the suggestions that it’s making before approving those actions to go forward and continue on. And so there’s very much an interest to go with the human in the loop.
But I believe over time, as there’s more trust in these AI systems and people learn to trust these systems. And honestly, I think the systems themselves will get better because we’re still in the early days of developing a lot of these tools, there’s going to be more trust and then there’s going to be more of a kind of autopilot, right? Where the idea is a person can just look at all of the suggestions and recommendations that an AI system has made and kind of look at them, agree, and press one button that says, “Okay, I’ve reviewed it, go ahead.” And I think that’s ultimately what is appealing for people. And so the AI systems will be more and more capable of doing useful things over time.
[0:19:30] DF: Yeah, it is crazy to see the impact it’s having today and knowing that we are still early days. These companies haven’t even had their IPOs yet and we’re just getting started on this.
[0:19:41] EL: There’s a saying that systems that we have today, this is the worst it’s going to be. It’s only going to get better from here. So it really does appeal to the imagination about how good these systems can become over time.
[0:19:54] DF: But any perspective on what the future two, three, five years out looks like with AI and how humans are operating with it?
[0:20:04] EL: Well, I definitely think that one of the things that is going to happen, and it’s been talked about in the past, but I can definitely see it, is that you will have the scenario where these systems, whether they’re from different companies, different platforms, they are going to work together in the sense that AI will work with other AI to get more sophisticated things done, and they will coordinate. I do continue to believe that humans are going to still play a role in just overseeing everything, making sure that things go the way that they’re supposed to. But you will have systems that are talking to each other as well.
For example, just one simple scenario in our case is like there’s a lot of AI tools around recruiting these days, right? That whole kind of front part of the process. But if a candidate gets far enough in that process, in the funnel, and it makes sense to turn it over to a background check system for either a pre-verification or a background check to kick that off. You can imagine that handoff being handled by an AI agent or at least suggested by an AI agent. And so that process could be smoothed over as well. So it’s all about just hiring and placing candidates as quickly as possible. Right? And so it gets back to the idea of even when there’s a team, a team of recruiters or a team of compliance specialists, just overseeing that process and making it go like autopilot.
[0:21:36] DF: And so you’re not on the – definitely going to help all of the processes. You’re not in the doomsday scenario, and mostly good coming from the AI. The Bill Gates camp?
[0:21:47] EL: You haven’t expressed what your viewpoint is with human-in-the-loop, so I’d be interested to hear. But yes, there’s always risk in everything that we do. But there’s just so much, I think, opportunity to design these tools in so many different domains that work well and do useful things. So we’re more focused on that perspective, that dimension, than the sort of frontier models that are going off and building super intelligence.
I happen to think that there’s just a lot of opportunity around building application or domain-specific AI that don’t need to involve super intelligence that are going to be way more useful, way cheaper to run because cost is becoming an issue with these frontier models. And so there’s going to be a lot of action in the next few years in these application-specific AIs.
[0:22:42] DF: Absolutely. And it does feel like it’s one of the best times ever to be starting a business or digging in. If you were a staffing owner today and you’re kind of trying to grow your agency, are there any things that you’re seeing when it comes to whether it’s use of AI automation or just kind of trends within the market? Are there things that you’re seeing that are working especially well versus other kinds of paths or strategies that you would kind of try to avoid today?
[0:23:07] EL: Well, there are a lot of tools coming out that are AI-based. So for the staffing agency owner, they must be bewildered by the amount of offerings that are out there. I think that I would say the same things as I’ve always said about tools and using technology. AI happens to be just another technology. It happens also to be just that much more powerful. And there’s some questions that get extra attention because AI is that much more powerful.
But definitely, it’s going to be the future. And I would just encourage staffing companies to look at experimenting with these tools and seeing where the value comes from because it is definitely, what technology and enablement is definitely going to be the way that staffing companies are going to get run. There’s going to be a large aspect of how you run operations with technology. And AI is the most powerful enabler of that ever.
So it’s worth it to understand how this works. And I think at the end of the day, though, it’s really important to make sure that there is ROI from the use of these tools, just like any other technology to make sure that you measure and see real ROI coming from these tools and to kind of take it from there. But I definitely think it’s going to be a meaningful space to look into and it’s worth investing into.
[0:24:32] DF: It’s funny. And coming off of Staffing World, a couple people I connected with, the conversation came up about, “Well, what is the ROI on AI?” And some people are like, “Oh, we’re not really seeing one. We’re not getting what we expect.” And then you have to dig deeper into those conversations. And it’s pretty much always, “Well, we gave our team ChatGPT.” It’s like, “What was the strategy? What was the plan? What process were you replacing? And how are you measuring what success looked like?” Which I think is one of the core elements of this.
One of the other things that this is going back a ways in our conversation, but I remember talking with you about the idea of candidates carrying their own verified credentials. Is that something that you still see happening? Is that the progress that you’re seeing with KarmaCheck today?
[0:25:13] EL: We’re still very interested in that idea. It definitely was one of the founding ideas that we had with KarmaCheck. And I was just recently at the PBSA Conference where that was a topic that continued to garner interest and discussions and so forth.
I think the thing that I’m excited about is that this kind of idea of wallets and so forth makes a lot of sense in really all industries related to background checks. It creates more efficiencies. It makes it more convenient for the candidate. It also makes it more convenient for the employer. It’s just a win-win all around. But what I’m most excited about is that in healthcare, the market that we’re serving, it really makes a lot of sense in particular, just because of the complexities involved in getting a candidate properly background checked, essentially, to get to the bedside.
And so where there’s complexity, the opportunity to use wallets, to use AI, this is just technology in general, is just the business case is really there. So we’ll continue to build towards that. We’ll have some more updates and developments, hopefully not too long from now.
[0:26:25] DF: Awesome. If I were an agency looking to buy back office AI, what are some of the questions that you would be asking to make sure that you’re making the right decision?
[0:26:34] EL: Yeah. First of all, just really making sure that the technology is really meaningful, right? It’s not just kind of a layer on top of ChatGPT or Anthropic. One thing that we’ve come to realize is really important with these AI technologies is just integrating into your existing systems, right? It has to be working within your workflows that you’re doing, your team is doing, and it’s not causing too much of an upheaval in those things, right? If the AI technology or the functionality can be integrated and work as part of what your team is used to, that’s really important. So we’ve seen the importance of that. So integration with your existing systems is, I think, really key.
[0:27:20] DF: Yeah, absolutely. And I feel like as we’re moving through the world, we’re starting to see more of the importance of security. And as each of these models roll out, how important that can be and how critical with B2B. Make sure you’re integrated and certified partner, marketplace partner, connected in a way that makes sense that you’re lined up with the account in the right way. Eric, any other closing comments for the audience today?
[0:27:40] EL: I think this is just an exciting time. There’s so much change, there’s so much progress. Really, I think the level of change and productivity coming from these new tools has just been amazing. I think it has surprised us how fast things are developing, where we were a year ago, where we’re going to be a year from now. It could be totally different. Everybody needs to keep their eyes open for what’s coming because it’s happening so fast.
[0:28:06] DF: Awesome. Well, this is Eric. Great connecting with you today. Thank you so much for joining us. And we’ll have you on the show again soon.
[0:28:12] EL: Yeah. Thank you, David. It’s been a pleasure.
