George LaRocque (00:08):

Hey everybody. Welcome back to Worktech. It's George LaRocque and we're continuing in our series meeting the Worktech AI innovators and I might have the innovator for AI and worktech on the pod today we've got Josh Swayne, the CMO from Paradox. The product is Olivia by paradox, and I was talking to Josh ahead of this, and I really think that paradox might've been the first AI forward solution to enter the market years ago. So in light of that, I'm really happy to have Josh here to talk about ai, how paradox views it, how all of this conversation around large language models and new interfaces, what paradox's view on that is and what the future might hold. Josh, welcome.

Josh Zywien (01:04):

Thanks, George. There's a lot of nice words about me and the company. Always been a good friend of paradoxes and appreciate the invite here.

George LaRocque (01:13):

Yeah, well, I love innovators and I love following paradox, so I'm excited to learn more today. Well, why don't we start, for those who don't know you and don't know Paradox, if you wanted to just do a quick intro of yourself and then maybe the business as well.

Josh Zywien (01:33):

Sure. I'll make the intro for myself quick, what people care less about, but I'm Josh Lane. I'm the chief marketing officer here at Paradox. I've been here for almost four years, which is crazy. And I've been in the industry for I guess eight plus because I was a SmashFly before that. Before SmashFly, I was with a venture capital firm that invested in HR tech startups, so I know the space well. And then I like to always say that my true expertise comes from my spouse who's TA and HR practitioner. She's with General Motors now, she's been with Chrysler in the past. So that's where I really learned the industry and I feel like the expertise comes from. In terms of paradox, we are a conversational recruiting platform. We've been around for, gosh, almost seven years. So we started at the end of 2016. I joined when we were 50 or so employees we're up over 600 now.

(02:24):

I think global offices, global clients. I think what we're most proud of is the fact that we work with some of the most amazing brands in the world and we've really seen their organizations transform. So we'll talk a little bit about some of their stories today, but our vision has always been that conversations could transform how we use software, how we interact with software, how we get stuff done in the hiring process. That was the vision from day one when Aaron founded the company. Obviously we're going to talk a little bit about chat, GBT and this move towards large language models and conversational as a interface, but it's been fun to watch that progress because I think we were banging the drum for conversational for a long time, and then open AI came along and broke those doors open, and I think now everybody's paying attention to it. So it's been fun to be part of the ride and I think most fun to play with or to help our clients transform their businesses and really see the impact of it. I think there's a lot of software out there that sounds nice, and then you implement it and you don't really see it move the needle. We've been fortunate to see it actually move the needle for some of our clients.

George LaRocque (03:32):

Yeah, that's great. I think it's important to step back and because of the way Chat GPT took over the conversation, it's almost used as a synonym for ai, but it's really not. It's one type of ai and we've been leveraging AI in HR tech and work tech for a long time, but it's behind the scenes and helping the experience, and maybe that's a great launch point for this conversation. Can you talk a little bit about how paradox has used AI with the conversational interface and what kind of value it adds for customers and what types of use cases or experiences candidates and recruiters and customers are having?

Josh Zywien (04:26):

Yeah, for sure. I think AI is still one of the more confusing things in our space. People get really excited about it. They tend to lump a bunch of stuff together and call it ai, and the truth is there's different flavors of it, and each of those flavors solve different problems or use cases. Ours really started with our founder, Aaron always tells a story of driving through McDonald's. He's looking on the window as he is about to order his food, and there's a little piece of paper stuck to the window with tape that says, apply to our jobs, and it kicked you to a career site and then you had to search for the job, then you had to find the job you want, and then you had to log in, create a login and password to actually apply to the job. Then that took another 20 minutes, and in that moment, his whole vision was, what if we could just do this over text message quickly?

(05:14):

By the time I got my food at the end of the drive-through, I'm scheduled for an interview. That was the hypothesis, could I do that? And the fun thing is we've done that. And so how does that work and what are the technical aspects of the AI there? It's really, I think we sometimes overdo it and overthink AI in this industry. We think that it has to be this mysterious black box that's got a bunch of algorithms and it requires a bunch of fancy PhDs from Silicon Valley. The truth is, if you just apply some basic automation, assistive intelligence, natural language processing to the different steps of the hiring process, you can actually remove a lot of the work that goes back and forth both for candidates and recruiters or hiring managers. So a couple examples of that. In most jobs, there's some basic screening that needs to happen, what are usually called minimum quals or minimum qualifications.

(06:07):

For every job, there's usually a recruiter who has to review every applicant that comes through and make sure that that person meets those minimum qualifications. So in trucking it could be do they have a license to drive over the road to cross state lines In nursing, it could be do they have a specific license or a number of years of experience in warehousing? It could be, are you old enough? Can you legally work in the United States? Can you live 50 pounds? All those things can be automated, so you can screen automatically for those requirements. And if somebody meets them, then you might want to just immediately schedule them for an interview. So in the case of a warehouse employee, you don't care much more about those minimum qualifications. If they have those, you want them to talk to somebody and maybe you want to offer them a job within 24 hours.

(06:54):

For nurses, it's more complicated. Obviously there's additional screen that needs to happen often in-person interviews, but you can also use AI to automate the scheduling process. So that's another part of the process where it's usually a huge time suck for everybody by the time the candidate gives you days and times that are available and you go back to the hiring manager and look at their calendar lines have crossed, and now all of a sudden that hiring manager is no longer free. So what used to take, in some cases, five days to schedule a single interview, we're shrinking that down to five minutes. And so if you can do that, you can see the impact that it has both on the experience for the candidate, they've gotten an interview a lot faster, they know what the next step in the process is. For the recruiter, it's hours saved per day, several hours per week.

(07:42):

It's oftentimes replacing the need for somebody to be dedicated to that too, so they can repurpose that employee for another part of the company. So it's things like that where I think it doesn't have to be this crazy technical, crazy mysterious thing. It's just looking at how you can apply technology to things that humans shouldn't be doing or don't need to be doing. I think that's kind of the thing that chat GBT has done too amazing. It's a great technology, but when you really look at it, it's taking these tasks that we used to do manually and simplifying them down to seconds and through a conversational experience.

George LaRocque (08:19):

Right, right. And when you say, you said the fun thing is you're doing that, I mean, I think it's important, literally you are doing that for McDonald's. I mean, that's not just a broad example, which is pretty cool. And I know I've spoken with folks in the industry when you were coming on the market and those were the stories that I was hearing. I tested this out, I was waiting for a table and I had an interview scheduled by the time I sat down before I sat down. And that's such a huge shift in the experience that most brands are providing. Now, those types of examples we talked about are frontline and high volume type examples. I think Paradox has also got some customers that are not just frontline or high volume, is that right?

Josh Zywien (09:19):

Yeah, absolutely. We kind of segment the business by the different use cases for our clients. And oftentimes if you're a large company like General Motors, you hire both. So it's not like you're only frontline or only high skill, but for General Motors, it might be less like you're trying to take over the entire hiring experience and make it all via text because that just might not make sense for an executive role or a software engineer or whatever. In those cases, you're just looking for the individual steps in the process that you can automate to make things simpler. So in the case of gm, it might just be interview scheduling or it might be the initial job search for the candidate. Job search typically stinks to go onto a career site. Let's just be frank, most career sites aren't great. If they kick to an a TS job search experience, it's usually even worse.

(10:08):

What we try to do is just bring that to a conversation. Somebody can come to the career site and say, in the case of General Motors, hey ev, I'm interested in a software engineering role in Santa Cruz, California. And EV will respond and say, that's great. We've got these five positions open. Do you want me to help you apply? That's a much better experience than what people are used to, which is starting with a login screen where they've got to either remember their username and password or create one, and then that creates all sorts of messiness. And then for the recruiter, it's just if you can automate interview scheduling and we do it through a TS integration. So in the case of Workday or SAP for example, it's just the status change. All the recruiter has to do is go in, change the status to schedule interview, and then our technology reaches out to, they're going to look at the hiring manager's calendar, and then it's going to reach out to the candidate and say, Hey, how does Tuesday at 2:00 PM work? And if the candidate says, fortunately, I'm not free, then we can propose new times. So it just takes that load off the recruiter's plate and again, outside of the hours saved and the time saved, it's just a better experience for everybody.

George LaRocque (11:16):

And speaking of experience, there's a branding element there that you were just mentioning. General Motors has a huge commitment to electric vehicles and it sounds like they've named their experience ev That didn't get past me. Yeah, yeah,

Josh Zywien (11:35):

Exactly. And then US Express, I think Jacob Kramer would be very happy that I'm sharing this because it was really his idea that started it. They actually have multiple assistants and multiple personas by different job types or different parts of the country or for their military hiring or veteran hiring. They'll have a different persona for that too. So it's kind of cool they can actually personalize and customize it based on the candidate or the type of job that they're hiring for. And it does for the candidate make it feel like more of a warm interaction or for gm, it makes it feel more on brand. So yeah, even McDonald's I think personalize their assistant with it's got McDonald's gear and the full uniform of a crew member.

George LaRocque (12:15):

Well, that's cool. So where is this all going for paradox? What are you thinking about working on? What can your customers expect thinking of AI holistically as well as chat GPT?

Josh Zywien (12:33):

Yeah, I think for us, the chat GPT stuff is just, it's clearly a wave at this point, and it's not just in HR and TA technology. Obviously we look around and I think it's most likely to disrupt software engineering and things like that very quickly. I've seen chat GBT write code, and I'm not a coder, but come from a family of engineers that is as good or better than a human could do. So it can debug that code automatically. So I think things like that, we're going to start to see major, major transformation in our industry. I think there's enough nuance and context necessary. You truly have to understand how the hiring process works. You can't just take broad general technology and apply it to our world. There are regulations that matter, there's integrations that matter. So there are things like that that I think it's going to be really hard to take a general technology like chat GBT and apply it to the complexity of how an organization like Pfizer hires or General Motors hires.

(13:33):

But that's where I think where we're well positioned is that we've believed forever that conversations and conversational will be the new ui. It'll just be how we interact with software. So now it's how we apply to jobs soon. It could be how do I access my payroll stub, how do I change my benefits? How do I change my 401k contribution? How do I share feedback after my first month on the job? So there's all these applications into the employee life cycle where I can see these conversational experiences, just simplifying things. So that's the way, I guess the lens that I would put on it is look at your current hiring process, your current employee experience process, where do you see friction today? And if you see friction, it can probably be addressed at some point, whether it's today or a year from now or five years from now by some sort of conversational experience that will simplify that dramatically.

(14:31):

So I think that's what we're really excited about is the number of use cases and applications for technology. This are massive. Our limitation is our ability to really decide where we want to focus effort and time. And thankfully our clients kind of push us in that direction. Oftentimes our clients will be banging on the door and saying, we really want to move you in this direction. And when we get collective mass there, then we go and we build it. But I think that's it for us is more this, I love that chat. EBT has raised awareness around the power of conversations and conversational now allowing people to imagine what that might look like applied to different problems.

George LaRocque (15:15):

But to your point, there are a lot of things that any customer or anyone thinking about how to leverage this technology should maybe take a beat and start to consider. And it's of course there's the broad security issue, but there's privacy, the use cases that you mentioned. There are external large language models and then you've got internal data. And in order to fuel the conversation, you have to rely on both. If I'm going to talk to someone about their balance of PTO or their benefit selections, I've got information that we don't necessarily want to educate the large language model externally. So is that something that you're spending time on? I'm assuming based on your comments it is. But what are some of those issues that you're thinking about?

Josh Zywien (16:12):

Yeah, for sure. I think our labs team, the moment that they could get access to the open AI APIs, they got it. They were in there, they were playing around with it. So there's a lot of stuff that we're prototyping and playing around with. I think we've taken a bit more of a long view approach to this of we're not going to be one of the first companies to rush out in the first six months and put out a press release and a new announcement. I think partially because we felt like we were already kind of there. The idea of conversational experiences wasn't new to us and we didn't feel like we just had to release another chat GPT thing. So that was one piece, but we are certainly playing around with it and seeing what use cases make sense. You mentioned compliance. I think that's one thing that companies should be very much aware of.

(16:59):

Anytime you're using an open API to another technology, then it introduces another layer of evaluation that somebody in the IT department, the CIO, the head of compliance and risk is going to have to evaluate that stuff. So I think that's a serious consideration because if it's not native to the vendor that you're using, you do have to consider the implications of that. So how confident are you that connection is secure? Where is that data being stored? How is it being used? Who can see it? As long as you can answer those questions and you feel okay about it, then I think there's a ton of potential with the GPT engine. If you don't, then you should tread lightly. And I think that's the approach we're taking is treading lightly, making sure we check all the right boxes. We work with some amazing clients that are huge and they have significant compliance risks and we're not just going to jump into the pool without knowing what's at the bottom. So I dunno if that answers the question, but I think that's exploring it, but also taking a bit more of a pragmatic approach to make sure that whatever we build will both solve a problem in a new way for our clients and that it doesn't introduce unnecessary risk that the client doesn't need to take on.

George LaRocque (18:18):

Yeah, I mean it makes a lot of sense. And I think listening to you talk, I think there's, there's sort of out of the box solutions that something like chat GPT solves, it's help me write a better email, help me shift it to a certain voice, et cetera. A lot of those tasks that take time, find me some examples of something out in the wild. But in order to really differentiate my employee experience, my candidate experience, I'm going to need to factor in my culture, my data. We all have access to chat GPT now. And if we're looking at putting this in our core worktech stack, the question should really be, what's the return back to the business? How can we really leverage this? Not just there's a huge advantage to writing things faster. There's an ROI on that, but that's already out that how left the barn or whatever the saying is. But I think you've got to take a longer view both to differentiate and on the compliance and security issues.

Josh Zywien (19:36):

Totally. Yeah. It's enough landmines out there that you just got to be careful where you're walking. That's it. We do believe that the large language models will help us train the knowledge base faster, so it'll be easier for us to develop to ingest documentation from a client that allows us to develop better, smarter, more adaptable answers to questions that candidates might have. So from a Q and a perspective, the potential here is huge. What we're trying to weigh is we can already do that. So it's not like we need the large language model or open AI to do it. It just makes it easier and faster for us. Does the risk of that outweigh the benefit? And so I think that's where we're trying to be, take our time a little bit more and be more thoughtful around what this is and what door we're walking through before we walk through it so that in six months or a year, if we find out that this is the better way to do it, we're ready to go and we're delivering a better solution for the client. If it's not and there's risk, then we're going to stay away from it. We're trying to protect our clients from that stuff too. And the alternative is we continue doing things the way we've always done them, and that's proven to be pretty successful.

George LaRocque (20:51):

Right? And we have had some starts and stops in other areas leveraging whether it's AI or different technologies that have been, whether they've introduced bias to the process or compliance risk, including bias. So there's a good reason, especially for the types of customers that paradox has to take that time. And if you're one of those customers, I would think I'm counting on Paradox to take that time and understand. I mean, we're hearing about now we're learning about what do they call hallucinations that the AI has. They just make stuff up and that's all part of the experiment. And with any technology, AI or not, it's implement, monitor, test, adapt, and I think that it's important. This wasn't launched when it was fully baked. So I'm excited about it. We're having these conversations because of it, but I think it's especially good advice. Anything else you wanted to cover before we wrap up or any other thoughts for buyers out there?

Josh Zywien (22:02):

No, I don't think so. I think it's confusing If I was a buyer out there, I work in technology and I feel like I have to read up every single day to stay on top of what's happening. So I empathize with our practitioner friends out there that are trying to, we see this with our clients. They're trying to solve a problem in a new innovative way. And for that, I applaud people because I think we've done it in this industry the same way for way too long. I joke with some of our clients that you could have gone full Rip Van Winkle and fallen asleep for 15 years and woken up today. It would still kind of look the same. It's not like a whole lot has changed. So for any practitioner that's out there trying to explore new technology and push the boundaries and try new things for that, that gets me really excited because it tells me that the culture shift and the mindset shift is really changing.

(22:59):

What I would just say is don't feel too much pressure to make a decision right now or to follow the trend just for the sake of following the trend. Don't start making a requirement in your RFPs that somebody has to have a large language model or something like that where you don't fully understand why that's being added to a requirements list. So take your time. Don't feel the pressure. There's plenty of technology out there that's not chat GBT, that's not a large language model that can solve your problem today. Map that out. Talk to experts like yourself, talk to other analysts out there. Try and do as much research with objective parties as you can. As much as I like it when clients come to us and ask for our advice, we're a vendor too. So go outside of the vendors that you're working with.

(23:48):

Try and get unbiased, third party views of things. Do as much reading as you can to truly understand it. And then be patient. It's all here. This is great. I think technology in the next year is going to explode and we're going to see all sorts of new use cases, but find a vendor that you trust. Make sure that vendor is positioned well for the future. They can build things for you in the next six to 12 to 18 months that are going to help you. And then make sure whatever technology you're buying is actually solving a real problem versus it looks and sounds sexy and it's something you can sell to your team internally.

George LaRocque (24:23):

Right, right. That's great advice. Well, Josh, it's always a pleasure. Thank you for being here. Thanks for informing everyone around Paradox and AI and where you're headed. Really appreciate your time.

Josh Zywien (24:37):

Yeah, it's a pleasure. And I'm sorry my dog was jumping up half the time. I don't know if you made a little sneak appearance, but

George LaRocque (24:42):

Thanks for having me. What makes this, it's the modern age, the modern era. We could have had mine here if I knew, but Well,

Josh Zywien (24:50):

Actually, I moved here too. My office is upstairs. I was trying to get to a place where I wouldn't be distracted by dogs running around in the background. And of course, you can't find a safe place anymore.

George LaRocque (24:58):

Right, right. Well thank you and thanks everyone for watching or listening and we'll, until next time.

Josh Zywien (25:05):

Thanks, George.

George LaRocque (25:06):

Thank you.

‍