Latest / Recruiting Evolution / Industry Spotlight | Aaron Elder - CEO at Crelate - How Early Adopter Recruiting Teams are Pulling Ahead with Agentic AI

Transcript

Industry Spotlight | Aaron Elder - CEO at Crelate - How Early Adopter Recruiting Teams are Pulling Ahead with Agentic AI · Recruiting Evolution

• Aaron Elder [00:00:00]: If you're not using some sort of AI in your daily life, I feel like you're being left. You're just not going to keep up, period. Right. So I think my advice would be think beyond, though, just the KPI wheel and don't confuse change of progress. That would be my advice. Are we just shifting the KPI wheel to have the word AI in front of it now? It's the AI KPI. We've coined a new term for you. All right. Aaron Elder [00:00:27]: I wouldn't focus too much on that. I would focus on where do you want to be two to three years from now as a company? How are you going to use AI to help you get there, but focus on selling those problems that you see in your industry that are two years out. Things are changing fast. All of the conversations you've had paired with your ability to go have new conversations is your only real secret sauce. Everyone has the same tools, and the tools are getting better and easier to get to all the time. And so our dreamboat, the Living Platform, is to really help you unlock the goal that is your only real proprietary asset and to let you deploy agents internally on your data to do it. I mean, that's the whole idea there. Kortney Harmon [00:01:05]: Hi, I'm Kortney Harmon, Director of Industry Relations at crelate. This is the Industry Spotlight, a series of the Full desk experience, a Crelate original podcast. In this series, we will talk with top leaders and influencers who are shaping the talent industry, shining a light on popular trends, the latest news, and the stories that laid the groundwork for their success. Welcome back to another episode of the Full Desk Experience Industry Spotlight. Welcome back to another episode of fde. I'm Kortney Harmon, and today we're back with crelate CEO Aaron Elder, and we're going to be exploring part two of our previous conversation where we're kind of talking about what Aaron calls this post AI era, when artificial intelligence has shifted from novelty to necessity overnight. So we're diving into how raising expectations for quality and speed are forcing every professional to rethink their value prop. So if you've been wondering whether your industry has crossed into this new paradigm or how to thrive in it, this episode's going to give you answers where the world and the rules are being rewritten in real time. Kortney Harmon [00:02:18]: So, Aaron, I want to revisit something you said last time. Thank you, first off, for spending your time to be with me again. Aaron Elder [00:02:24]: Always fun, Kortney. Kortney Harmon [00:02:26]: I love it. Okay, so you had mentioned last time we talked that you've been intentionally using the phrase post AI era. Can you break down what you mean by that and why the terminology matters in this instance? Aaron Elder [00:02:39]: Sure. My intention with that was to try to evoke some sense of urgency and emotion and acceptance with it. It's not really a question or a debate whether AI is impacting things. I mean, this is. It is, it's. Your clients are using it, your customers are using it, your people are using it, whether you know it or not. And so it's really about acceptance of that and then forcing yourself to navigate it and take advantage of it. Kortney Harmon [00:03:04]: So it's no longer in the wings, it's here. We're using it. So post AI, we're over that hump. Aaron Elder [00:03:09]: Correct. And I think the thing that caught me personally off guard, and I mentioned this last time, was that the rate of change was. You know, the funny thing about like exponents and rapid growth is that it compounds. And so you saw AI two years ago, it's like, oh, that's interesting. And last year it's like, oh, it's getting kind of cute, et cetera. And then all of a sudden it's everywhere. Right? Kortney Harmon [00:03:30]: Yeah. Aaron Elder [00:03:31]: And that really just happened in the last year. Right. I mean, the LLMs were really like last summer sort of crossing that uncanny valley in a way that everyone could use it. And it started see real value. Kortney Harmon [00:03:42]: And I think you see this differently. You see what is going on, how things are changing, how our products are developing. But like you mentioned, AI went from a toy to essential overnight for someone really just waking up to this reality. Because there are people that are just waking up to the reality of I'm behind on AI, I need to be changing. What are those telltale signs that we've crossed this threshold? Aaron Elder [00:04:06]: We cross a threshold. Well, the funny thing is that I'd be willing to bet that most listeners are engaging with AI at least once a week, if not daily, without even knowing it. Because companies are deploying this stuff so rapidly. You know, you're getting emails in your inbox. They seem personal, but they're probably generated. You're calling an automated support line, you're talking to a help desk or a support system. The first line is awesome. Often going to be AI driven. Aaron Elder [00:04:31]: Companies just can't deploy this stuff fast enough. And it's touching sort of everyone's experiences already. I don't know if that answers your question, but I mean, it's, it's everywhere. Kortney Harmon [00:04:39]: No, just like what made it go from a toy to essential? Yes, everyone's using it, but what was that threshold like? When did that Change. We went from LLMs to. I just was at Colorado Staffing association virtual conference, and we talked about agents and everyone's like, oh, I don't know what's going on. Like, it's change. Aaron Elder [00:04:56]: Well, I think a few things happened. One, the large LLMs got good quickly and it sort of crossed. So. So llama. So Facebook's llama3 kind of came out, and that was kind of a big thing. And what it did is it, from my point of view, sort of sparked the open source. Not open source, but the open model race. And so not only was it a really good model, but you could host it yourself without having to Pay Facebook, unlike ChatGPT, where it's behind like a paywall and it's all owned, there was sort of this race to like, it's really good and it's free, and then march along a little bit, deepseek came out and that kind of like blew everyone away. Aaron Elder [00:05:37]: Because now not only was it good and free, it was cheaper, a lot cheaper to host. So the quality was maintained while the speed of what's called inference went up. And so now everyone can have access to it and they can host it, and it's pretty good. And so, I mean, you can see how adoption would scream, right, with that scenario. Kortney Harmon [00:06:01]: It's not just incremental progress anymore like it is overnight. Huge shifts. Aaron Elder [00:06:05]: Yeah. And I don't remember the exact savings, but, you know, I think deep seeking could be hosted for 80% less than, you know, the previous versions could be hosted. And internally they had this pretty kind of novel idea of creating like a collection of experts. And so instead of having one big model that had to know everything, they started breaking into like smaller little mini models that were each trained for an area. And then it goes to the first model to then go to the next one. And it's all about like, these models are massive, like billions and billions of parameters. You know, you know, good ones are 70 billion plus. And you got to have all that in memory in order to be able to use it. Aaron Elder [00:06:43]: DC kind of broke that down, said, no, you don't really need to have it all memory. There's also other advancements that are happening. Microsoft is working on a one bit model. So most models, each data point, I'm sorry, I get a little technical, is stored as a floating point value and it has a certain amount of precision to it that takes more memory. You can distill all the way down to just a single bit and still get pretty good results. And you can host this thing on your phone and it produces text pretty good. Pretty darn well. Makes sense. Aaron Elder [00:07:12]: And so you can see as the cost of these models go down, down, down, down, the usage will be everywhere. I mean, yeah, I love it. Kortney Harmon [00:07:20]: That's way more technical than my brain can comprehend. That's okay. I love it and I love that you understand that. So I mean, there's so much to this and obviously we've seen beyond the obvious automations. Right. What is the second and third order effects that you're seeing? Because AI is really becoming universal in business operations. Aaron Elder [00:07:39]: Well, and actually answer part of your first question too on that. Because the models is only so much. It is the generative capability, if you will. But what I think everyone saw with the first sort of iteration, oh, I'll just log into ChatGPT or Copilot, I'll type my thing, I'll get my result and paste. Right. And so what has happened in parallel to this is the agentic side, which is where you wrap, you make the models accessible and integrated into your workflow with your context and even give them autonomy to go take action on your behalf. And so those sort of two trains were happening in parallel. And it's really interesting that when we, I mean, we're in the process of rearchitecting Crate to be an agentic first platform, and it's a different way of thinking. Aaron Elder [00:08:26]: It's very similar in many ways to microservices. Again, sorry for getting technical, but it's different in that you're creating these agents that make multiple calls, 20, 30, 50 calls to the models as it works through all the logic and patterns and engages with you along the way. And so that's way more complex than just ChatGPT copy paste, if that makes sense. Kortney Harmon [00:08:53]: No, that makes complete sense. Aaron Elder [00:08:54]: And so what this has done is this has allowed platforms like us to think about how can we integrate this into pretty much every single thing we do. I can riff more on that too, but go ahead. Kortney Harmon [00:09:05]: Yeah, I'm all ears. Because I'm going to continue to ask more questions about it. So if you want to continue, great. Aaron Elder [00:09:10]: Well, the way we're thinking about it is we're thinking about it in terms of agents being a user experience first. Okay, so you are engaging with the platform via an agent, right, or an assistant, kind of depending on what the thing's doing for you. And when you think about that, I mean, there's a world a couple years from now where you'll never log into Correlate directly. Right. Or a web browser for that matter. You'll just engage with the agent, right, as your primary interface. I mean, that's kind of the direction things are going. For now, though, we're thinking about like, Yeah, I mean, the experience is I want to engage with my agent first. Aaron Elder [00:09:45]: I want to know what the agent's been working on, what's been doing. I want to be able to talk to it. I want to be able to naturally chat with it to get it to change those results. But I also want to maintain control and be able to see what it's doing when it's doing it and, you know, decide what happens next, control over it. Because ultimately, you know, we are in a people business and I think it's absolutely critical to maintain. You know, during this whole process, you. Kortney Harmon [00:10:05]: Kind of were going down this track just when you were talking. But I know that our previous conversations, you said businesses really need to think if they're even strategically asking the right questions and solving the right problems. Right. And in this post AI world, how should leaders really fundamentally rethink their business models and strategies when it comes to AI, Are they creating or are they asking the right questions, but are they solving the right problems and how they move their business? Like, how should they fundamentally rethink how their business functions today? Aaron Elder [00:10:38]: I don't know if I can answer that for everybody. There's a wide, I mean, what we're thinking about, again, it's sort of a layered approach. One, you can't assume that the AI can do everything out of the gate, even with the great context and all the agents, all the things, your mileage may vary and the more narrow the task, the happier you're going to be, I guess, is my sort of general thought or experience at least. Right? Kortney Harmon [00:11:03]: Yeah. Aaron Elder [00:11:03]: You know, it's really hard to train an agent or build an agent that can just do everything for everybody. You have to hone it. And so the more narrow of a task that you can think of target, the better off you're going to be. Now, within that narrow space, you might expect it to do a lot of wide range of things or generative responses, et cetera, and that's okay. So I think it's minute number one is being smart about where you actually want to go apply it. Two, you got to try a bunch of things, you know, at correlate. You know, we've been trying AI for support and for development and for sales and for marketing, and each team is trying different things to sort of see what works for them in their, in their use case, et cetera. We've deployed an AI agent for all of our developers. Aaron Elder [00:11:46]: Right. So every developer now uses an AI agent, you know, to help code. The adoption of that has been ridiculous. But now you get into sort of the user point of view. Users have to start learning how to not only manage within this narrow task, but also to sort of not lose their humanity and, you know, the brains, the operations. Here's a reality. If your job is just type in something to a agent and spit it out to your boss, well, then we don't really need that person very long. Right. Aaron Elder [00:12:14]: I saw this MIT study which showed people's cognitive functions, like, rapidly decline with the rapid use of AI. Like, you can really just stop thinking, which then. So this is my third. My third point of this is that you actually have to think more, I think. And I think if you can think more, holy cow, productivity can go up a lot. Kortney Harmon [00:12:33]: Yeah. Aaron Elder [00:12:34]: Super exciting. I mean, those are the day where I was working on. I was coding three different projects simultaneously. I had the agents helping with each one, kind of all in parallel. And it was cool because I could sort of start a train of thought and let it go. And start a train of thought and let it go. And then I could solve this problem. But you have to stay engaged and thoughtful at each level. Aaron Elder [00:12:53]: Because I've heard this a few times from people I've experienced myself. AI will very confidently tell you the wrong answer. Kortney Harmon [00:12:59]: Yeah. Aaron Elder [00:13:00]: Like, straight up. And then, you know, you're like, hey, is this the best way to do this? And it'll be like, yes. And you're like, what about blank? And it's like, oh, that's way better. You should do that instead. Kortney Harmon [00:13:12]: Best cheerleader around, Right? Aaron Elder [00:13:14]: It's almost a cheerleader. And I've seen, you know, you ask it to write a job summary or. Or a propos. You'll present a candidate to a client and, like, it'll just make up stuff about the candidate, right? Yeah, because it inferred it from whatever. So you got to stay in the. In the mix. Kortney Harmon [00:13:29]: I love it. I might get smacked from the people backstage by asking this. So we might have to take it out if we have to. But whenever these organizations that we're serving that are our clients are trying to fundamentally rethink their business model, how can our agents that maybe we've done today or you're developing help those people rethink their business? Like, you're. You're like, coding three things at once. How are we helping this industry do the same thing simultaneously for their audience? Aaron Elder [00:13:59]: Well, so we are Taking, I mean, like I said, a fully agentic approach to this. And we're thinking about future user interspaces and user experiences from an agent point of view. And so we're building agents that are targeted towards certain parts of the recruiting hiring process. And so again, back to that narrow tasking. The more narrow a task, the better we can train it, the smarter we can make it. And the truth behind the scenes of agents is that, and I don't want to say it's smoke and mirrors, but between those 30 or 50 calls to LLMs is human driven logic that we have built into the agent to make it dubious things. And both the LLMs and the agents and that logic all have to get better over time. And so there's a thing called observation or whatever, you know, observability. Aaron Elder [00:14:49]: And so like we're measuring every single step along the way so then as it gets used we can tweak it and make it better and better. Man, that's a long winded way to get back to your question. So we're targeting the agents at different purposes in the business. Like the thing you're trying to do is go find new clients, go find new business, go find new people. That's not in your database. We're trying to build an agent specifically for this sort of sourcing and business development kind of motion. Second, if you're trying to unlock the proprietary data that's in your system, your secret sauce, all your hard work, we got a dedicated agent for that as the insights agent. And they share some similarities for sure, but they're definitely targeted and purpose built for very different functions and they work very differently after that. Aaron Elder [00:15:31]: We're looking at an operations agent which is going to be really focused on helping you run your business, keep your data clean, all those kind of things. And so we're thinking about it in terms of end user function and what the agent's trying to help someone do. So presumably as you think your business, you're thinking, well, which people are going to be engaging with agents? How do these aligned with my roles? And then with that, how can those people become more effective? Right, you don't necessarily need the Boolean ninja anymore to go write all the searches for people. Like everyone can be a good source right now as an example, if it. Kortney Harmon [00:16:05]: Makes any sense, it makes complete sense. You hear so often like, oh, the robots are going to replace us. I don't believe that. I truly think that we're going to be upskilled in all of our roles. You mentioned we're Moving faster. The expectations are going to be different. I don't think whole roles will be replaced, but whole roles will be evolved. Aaron Elder [00:16:23]: I don't know, part of me we'll have to think that. But I definitely think some roles will be replaced or certainly absolutely some parts of them especially. Yes. And I think as you as an individual, if you're not on, I mean, I've done a thought to a wide number of people and like some people are like, oh, it'll never affect my industry. Right, okay, you know, wrong. It probably already is and someone else is taking advantage of it and so you don't. It's kind of that whole like old guard, sit on your laurels kind of thing. That's not going to work. Kortney Harmon [00:16:52]: No, no, I agree. But it will evolve who we are in our roles today. So. Aaron Elder [00:16:56]: And actually one of the points too is that I heard this, I won't say their name, but it was a sort of a, I tried AI, it can never do what I do kind of thing. You know, and my thought was one, it's sort of layer of thoughts, but like first off, you just clicking around. ChatGPT is not trying AI, so there are experts who are intentioned on using AI to solve that problem. They're going to figure it out. Second off, the experience that you had three months ago or six months ago or a year ago, like, or today, like that's the worst it's ever going to be. It's going to get a lot better. And so you really got to sort of be future thinking on both of those. I think the other thing that I've seen is people say like, oh, we tried it. Aaron Elder [00:17:41]: I won't say who it was, but they had tried a BD motion with AI and they're like, oh, that's terrible. I'm way better at blah blah, blah. And it turns out, long story short, it was the cheapest of cheap AI solutions. For the beauty thing, it was like 50 bucks a month, no implementation, self service experience. Like, okay, great. Meanwhile there's some other company that went out and got world class agent, paid a hundred thousand dollars to get that thing personally customized for their organization. Believe me, that thing's probably working. Kortney Harmon [00:18:10]: Oh yeah, right. Aaron Elder [00:18:11]: And so I think there's maybe this cognitive dissonance going on where people are just not accepting this or they're, they're looking too low when they need to be thinking sort of higher. Kortney Harmon [00:18:20]: I agree wholeheartedly. You had also made a comment at some point in time. I don't know if it was on the last podcast. But you said there's a difference between activity and progress. Yeah, but with AI raising this bar for both quality and volume in the recruiting industry. Talk to me about what you think executives should be focusing on producing to stay relevant. Aaron Elder [00:18:41]: Producing? You mean content wise? Kortney Harmon [00:18:43]: Or what are you thinking to their clients themselves? Like what do they need to be providing or even any business. It doesn't necessarily have to be recruiting, but I'm thinking in our space for direct hire offices, corporate in house, if what do they have to produce to stay relevant? Because I'm going to tell you, I've had some talks, I'm actually speaking at Colorado Staffing association in August about the KPI hamster wheel. You know, we always talk about like this, so many calls to so many this to so many presentations to so many interviews to so many placements. AI has thrown that on its head. It is no longer the same ratio reports that grandma and grandpa had whenever we started our firm. And I think the concept of what we're thinking of and what we're producing for our clients and our candidates has completely evolved. It's not the same anymore. Aaron Elder [00:19:27]: Yeah. Where my thought process goes is on focusing on outcomes, not necessarily the inputs and the steps along the way. Right. I love that term, the KPI hamster wheel. And it's funny when you look at early AI or even automation, it was really chasing that hamster wheel. Just can we make it spin faster? Right. Assuming that if we did, it would produce the outcome. Where I'm interested in positioning correlate and where I think our best customers will be wanting to position their businesses as well is helping companies solve strategic recruiting problems. Aaron Elder [00:20:01]: The world is changing. There's tariffs and there's macro stuff going on and there's AI. And companies are going to need to respond. They're going to see it as an existential need to respond. They don't have the teams to do it. They don't know what, they don't necessarily know what the outcome is or the questions ask. And there's going to be a rise of these like strategic recruiting problems that are going to happen. And I think it's higher value stuff than maybe what we've seen before because it's not going to be just throw bodies at it. Aaron Elder [00:20:31]: It's going to be a different mot. And so we want, you know, our customers need to be thinking, how do I position myself to solve those problems and have that kind of partnership with my client? It's very similar to what people have done in the past. But I I think it acknowledges this race at a bottom isn't going to work and that if you want to differentiate, you've got to be having those conversations and furthermore, unlocking your database. I mean, here's a funny thing, like everyone will have AI in a year. Everyone will have sourcing tools and access to LinkedIn and all the things. So other than your hustle, what's your differentiation? It's gotta be who you know in your database and your ability to connect with people and your ability to have mature, strategic recruiting conversations with clients that help solve that holistic problem. That's my thought anyway. Kortney Harmon [00:21:22]: I think it's great. We don't even know the conversations and the problems that are going to arise two years from now. Aaron Elder [00:21:29]: We don't know the problems. But imagine a scenario where like, you're trying to move your factory from China to Texas, right? To take advantage of changing things that are going on. Maybe you're a company that over, over time had only been designing. I'm talking big stuff here, only designing things that you've been shipping it all to China. And like, you don't own any manufacturers, a company itself, right? Like when it's the funny thing, like Apple doesn't make their own screens or their own chips or pretty much anything. They're. They're really a design company, right? And so you're thinking these things and so then you're thinking, like, man, we gotta go reinvent ourselves and all we have are, you know, lawyers that can push paper around or like, like, we gotta go build a whole new team. And. Aaron Elder [00:22:10]: And so I don't think what they're thinking is you gotta pound the phones and make 50 calls a day. You see what I'm saying? Like they're thinking like, okay, we have a wide range of talent problems that we gotta solve to solve this. Plugging into that is pretty key. That's where my thought is. And I think that works at many levels, not just the biggest of the big. So much bigger picture. Kortney Harmon [00:22:32]: No, no, I love it. It's a much bigger picture. I think it's great. Okay, we talked about AI is becoming standard, obviously in our toolkit. How are clients and stakeholder expectations for quality, speed and volume shifting and what does that mean for our structure of work? Aaron Elder [00:22:48]: My thought would be like, oh, you know, expectations are increasing, but expectations have always been high. Like. Right. Like one is a client's expectations not been high. Right. So I don't know if from that point of view, I personally have seen any change. I don't know if you've seen change or the experience is always high. It probably depends on if you're working internally or if, you know, if you're working at an agency. Aaron Elder [00:23:10]: As you know, the answer to that. Kortney Harmon [00:23:12]: Question, I think probably in the staffing industry, quicker is, I mean, speed has always been the name of the game. But more qualified, quicker, faster at more scale. I am sure at some point in time it's going to be the idea of how this evolves in the idea that it's, it's no longer two placements a week, it is six placements a week. Because our scale of what we're being able to present is faster. More qualified data is more at our fingertips. What do you think? Aaron Elder [00:23:42]: Maybe where that got my thoughts going was that might be a transitive thing, if you will. Right. Because, you know, people are concerned, they want, take advantage of things. They want to move fast. Move fast, move fast. Great, great, great. But maybe we got to think out a little bit further than that. And I'm thinking on two fronts. Aaron Elder [00:23:58]: Okay. One, I don't know if you guys or if you saw the headlines, but it was like, I think it was Meta was offering like a hundred million dollar pay package to take top AI talent away from key competitors. It was, it was something ridiculous. It was like $100 million option package. And I think Meta just pulled like the head of LLMs from Apple over and Apple's been really struggling with LLMs, et cetera. And so that is a very telling notion that the need to be able to make meaningful moves in a post AI world is freaking critical. I mean, if Facebook just wanted to do that and all the fact check that I don't really have the article, but it was something ridiculous like that on Hacker News. And so that's one end. Aaron Elder [00:24:42]: On the other end, speed, speed, speed is great, but you should go check out the latest, you know, Boston Dynamics and other and the stuff that Tesla's doing with robotics. Right. I really, truly do think that these humanoid robots with all the power of AI we have today are going to materially change large portions of labor in the future in ways that I don't think people are really prepared for. And so then the question is, is that, does your customer say you're a staffing firm? See you. And I know I'm talking a little bit of pie in the sky here, but like, do they see you just as a body shop provider or are you a managed service provider? And you know that as a managed service that you're going to be bringing in automation when it Makes sense. And that you will be. You're at least thinking about these things so that they don't have to make sense. Kortney Harmon [00:25:29]: Yeah, no, that makes sense. Aaron Elder [00:25:30]: And I guess that may. That's much more I thought GOES is responding today with speed and AI is great, but how are you thinking about providing this managed service, this holistic service, strategic, strategic talent solutions over the next two years? Sorry. Kortney Harmon [00:25:45]: Well, no, you're not wrong. It's really the fundamental shift of a value proposition. It's a change, right? And I think those rise of expectations and the output quality does change the value proposition that people haven't even thought about, Right. Aaron Elder [00:25:59]: I mean, if you are in the business of supplying taxi drivers and Waymo arrives in your city and it can deploy 4,000 vehicles overnight, that doesn't really help you, Right. Like, you need a different business model. And so I think you should at least be thinking about that. And I think if people aren't even thinking about robotic automation, then they're doing themselves a disservice. Just at least think about it, right? How would we take advantage of this? How would we provide value in that world? Kortney Harmon [00:26:25]: No, I like it. I'm gonna shift gears and come more back into recruiting specifically. We've seen thousands of professionals navigating this shift. What patterns are you seeing with the firms that are thriving for the firms that are being left behind in this AI instance right now? Aaron Elder [00:26:43]: Well, it's funny, the ones that just popped in my head of people who are thriving the most are ones who are reacting to what's hot. And when it's hot, they were doing developers and they shifted over to do an accounting, things are shifting. And so I think that's not new, right? So I think that's that. I think what's interesting is that AI can help accelerate that or at least de risk it, right? Because now you have access to sourcing and data that you didn't. And you have access to it in a way via an agentic experience that sort of lowers the barrier of entry for your people to start adding value in that space. It's just kind of where I know some firms were like, you know, there's definitely this theory of, like, Marc, you know, sorry, market mastery, right? Like, you know, like, da, da, da. But also at some level, you know, that only goes so far and only scales so much. And on the other end of the spectrum, professional recruiter that is good at their craft can go recruit anybody. Aaron Elder [00:27:35]: Right? And I truly see both worlds work successfully. But my thought is, is that at least at scale, you Know, shifting with the market and using AI to help you do that is probably the single biggest thing I've seen. Kortney Harmon [00:27:47]: And this isn't. I'm going to say this just because so many of us get like this shiny object syndrome. It's not a shiny object, it's a shifting tide. It's not the idea of it's one thing right here, right now. So I like the way you explained that. Aaron Elder [00:28:00]: It's funny, like internally we're testing the AI and the sourcing side of the things. And the testers are like, well, I was looking for a carpenter, but it was bringing back, you know, people who had worked at SpaceX and NASA and people are like, oh, that's so stupid, right? But you go look in and well, NASA actually employs carpenters, right? They have facilities, maintenance, they have, you know, good grounds to be done. They do models and like, like, like, who knew, right? And so I guess my thought there is, is that when you are pivoting your company into a new industry, you know, let's say the accounting side, your sourcers aren't all CPAs. Like, they don't know all this stuff. And so the agent does. And so it goes beyond just like semantic search. It's bringing up things that are relevant than what part of my team thought was a defect, was actually a feature, if you will. And so it's bringing up people that you may not have considered or, or you didn't know to ask because you don't know the industry. Aaron Elder [00:28:51]: Right. Kortney Harmon [00:28:52]: You didn't know that that role existed. You want to know a funny story, Aaron? I actually did my student teaching at NASA in Cleveland. I was a preschool teacher at NASA because they had a daycare. But who knew teachers existed at NASA, Right, Correct. For daycare. Aaron Elder [00:29:09]: But the agent knows, so. Kortney Harmon [00:29:10]: But the agent is picking up stuff that is putting together dots that we're not necessarily thinking of on the surface. Aaron Elder [00:29:16]: And so that then accelerates your time to value or democratizes the ability for you to add value within your company, within your existing resources, a new target space. Right? Which was, I think is super cool. Kortney Harmon [00:29:26]: I love that you mentioned a little bit more about kind of what we're developing or what you're developing here. And you're building a technology that learns and evolves. Bottom of the line here. Your concept of the living platform, right. What does this mean for society when our tools become genuinely intelligent partners rather than just strategic instruments? Aaron Elder [00:29:47]: Well, here's an interesting question sort of baked into that, like, do they become truly intelligent? Right. Like, like there's this whole like really race for general artificial intelligence. And man, you read the article six weeks ago and it's a foregone conclusion. It's going to be here before you know it. It's like, you know, it's, it's 18 months away. People are making predictions and then the last round of articles I was reading, like the last two weeks, never going to happen. It's, it's a, it's a pipe dream like, like people's expectations have gotten too far, you know, and I forget what it was. There was, there was one article that broke down. Aaron Elder [00:30:22]: Oh man. Who did it? I think it was Apple. Apple did it. And it was, you could argue, kind of in defense of, of their lack of progress on LLMs maybe is what people, some people would argue. But it was a pretty detailed breakdown about how LLMs are largely like a big bag of tricks and that there's not actually any real intelligence behind the scenes, like whatsoever. And so it was a, it was kind of fascinating like PhD level article. So sorry to unpack all that, like, do we actually hit these autonomous partners in crime, if you will. Maybe another question would be, does it matter if it's intelligent? If it's intelligent enough to do the tasks that, you know, again, back to the narrow task and maybe just get a little wider to support you and your overall value function? Like, again, I don't think it matters. Aaron Elder [00:31:10]: I feel like I lost a question in there. Kortney Harmon [00:31:14]: All I asked was, what does it mean for us in our industry or society or whatever that when they become a different concept? Right? Like, what does it mean for us as we start thinking differently? Aaron Elder [00:31:27]: I think it's. And I'll, I'll use the devs on our team as a, you know, an analogy. Like we went from. It's garbage. I don't want to use it to. Every single developer uses it every single day on every single project. And that happened in the span of six weeks. Right. Aaron Elder [00:31:44]: And so I think what you're going to see is you're going to see every function in the company, you know, in a, you know, every talent recruiting function will be using it every single day at some level. And so when, you know, actually goes back to how we're sort of thinking about our agents, like, like we're not pricing them per seat. Right? Makes sense. Like because you're, you're buying the agent and it's really about how much work the agent's doing. But just like any coworker, everyone has access to the work that the agents doing. If that makes Sense. Right. So it's a, it's a different way of thinking about it because I think everyone needs to be engaging with these agents and, and as they become more autonomous and as they become sort of working alongside you, I think you got to lean into that. Aaron Elder [00:32:24]: Right. Everyone needs to have access to it, everyone needs to be able to, you know, engage with it. Kortney Harmon [00:32:28]: Yeah. And it's going to be an everyday thing. And I honestly, I had a conversation with Lauren Jednet last week and we were talking about agents and the ROI behind agents and she told me more about agents than I didn't know. And certain agents only work four hours a week. You know, what we're developing, how that's different and it's working with your data non stop versus I didn't know there was a stop on some of these. So it has been eye opening to even have conversations with you and Lauren and everyone else on our team to understand how those pieces work. Aaron Elder [00:32:58]: Definitely. Kortney Harmon [00:32:59]: I love it. Obviously AI is going to be handling more cognitive work. What becomes maybe uniquely irreplaceable, how does that work with humans? Because let's face it, we're in a people business and it's doing more things. People are worried that it's going to replace the human element. You've already said that it needs to work beside you. You know, how do we think about them? How do we think about AI thinking more cognitively replacing humans in that professional context? Aaron Elder [00:33:27]: Well, certainly in our industry, legally, the actual decision making around hiring has to be left to a human. And we've already talked about this, but there's just laws that are working their way through. You want to be able to be the person that made the final decision. And if the AI helped, you got to be able to defend how the AI helped and how it arrived where you arrived. And so we, we take that very seriously. And then so like our agent approach to that front is one, showing you you know what it's doing and two, keeping you under control and out of a nature. We need to, well, I mean we need to design it as such, I guess is maybe what is how I was getting done at Thoughtcraft. Right. Aaron Elder [00:34:02]: You have to design it as such. Kortney Harmon [00:34:04]: Give me more on that. Aaron Elder [00:34:05]: Well, one, what, what I was saying. Right. But two, so one, the system itself needs to not create the expectation that it's going to hire for you. Right. Like full stop. Two, I think that the team needs to be aware that again, like if you're offloading your cognitive functions on onto the AI, arguably like any other muscle in Your body is going to, you know, your own brain will atrophy. So, like, you have to stay engaged almost as a, as a rigor of practice if, if you will, in what it's doing. And it's kind of funny. Aaron Elder [00:34:37]: I'm hoping as we deploy our agents, that's gonna be more of a collaborative effort with our customers. Right. So we're, so we're tracking, you know, what it's doing, but I'm hoping we're gonna get a lot of feedback and helping, you know, make this thing smarter and smarter and smarter. My hope would be behind this guy. My hope would be that this offloading of the cognitive abilities is actually sparking or unlocking you to have new, better ideas, if that makes sense. You type in a Boolean search and finding some stuff that's not really heavy lifting, you seeing the diamond in the rough, that maybe the idea got sparked because the AI found this person that you had talked to before in an unrelated way that you didn't realize. Okay, now you put two and two together and now you make something happen. Yeah, that would be my dream. Aaron Elder [00:35:22]: As to what it would have, I think that's great. Kortney Harmon [00:35:24]: Do you think these AI agents, like, even what we're creating here, that's the evolution right now, today. What do you think's next? Is it agents talking to agents? Your agent talks to my agent. What do you think is next beyond that? And I know you don't have a crystal ball, but I would love to know your opinion. Aaron Elder [00:35:41]: Right, so, so what's next? The absolute next thing will be agents taking more action autonomously. That there's this whole thing called model context protocols, which basically is a framework that allows developers to go do that stuff. Right? So, like, those frameworks are evolving quickly, but they're pretty, pretty freaking new. Those things are going to be everywhere. So right now they're kind of at toy stage. Like, just maybe you've tried one of them out where you can install it on your desktop and it'll like browse Amazon for you and buy stuff. Like, I mean, it can do all that stuff, right? The next evolution of that is that it has a direct connection with all your systems. It can talk to Curly, it can talk to the web, it can talk to your marketing system, it can talk to your accounting package, it can talk to whatever you, you allow it to. Aaron Elder [00:36:20]: So I think that's probably the next big wave is this, this direct connection to different systems from the agent. So then the agent experience lets you go do something. So you can say chatgpt book me a reservation, you know, at the best restaurant in Boise. ChatGPT connects to your OpenTable account and does that for you make sense? Kortney Harmon [00:36:43]: And yeah, it makes total sense. Aaron Elder [00:36:44]: And in an ideal world, they didn't have to go purpose build that connection. The agent just knew how to use it intuitively by virtue of these connections. And so as everyone opens themselves up to this, I think those actions will take off like wildfire. Kortney Harmon [00:37:01]: And do you think that will happen in agents? Like what we have? Like, the insight agent will talk to the discovery agent, then we'll then talk to the BD agent. Aaron Elder [00:37:08]: Oh, yeah, we're already doing that. And not just the agents talking to each other, but then the, you know, actually go and modify the data. Like, oh, hey, I noticed that this looks wrong. This looks out of date, you know, based on your permissions, I just went ahead and updated it for you. Right. Kortney Harmon [00:37:21]: Love that. Aaron Elder [00:37:22]: Fast forwarding past all that. Does everyone have their own personal agent? I think that was a question that you sort of baked in there, like, and then like, do agents talk to each other? Like, does my agent call your agent? And we figure out, like, that's just kind of weird at some level. Like, how many levels deep do you go? Kortney Harmon [00:37:37]: I don't know. But I would love someone to help me do all the things that I have on my plate. Aaron Elder [00:37:41]: Here's an interesting thing that hasn't, and I just posted this on Facebook, is that the thing that really powers all these agents and yellow lens behind them are the ability to ingest large amounts of data for free. Right. And Cloudflare did this. Cloudflare, I don't know what it is. A third of the world's Internet traffic runs through a Cloudflare server. So they see a lot and they basically analyze and figured out that these LLMs take way more than they give, right? Like they do not send traffic back to the site. They consume way more often. Right? So a normal web crawler comes around, you know, once a week or whatever. Aaron Elder [00:38:17]: These guys are multiple hundred times a day and they return back a fraction of the traffic. Where am I going with this? Is that as you start building these agents, there's sort of two maybe big problems that have yet to be solved in the sort of post AI world. One is what do you do about the content? Right? So as content producers are no longer producing because they're no longer getting paid, now the knowledge cliff of LLMs gets stale because there's not a lot of new knowledge being put out on the Internet or behind, you know, out There free. Second, that knowledge is being sort of perverted, if you will. If agents are building all this content, if 4 out of 5 blog posts are AI generated a year from now and LLMs are training themselves on new blog posts that are largely generated by like what exactly is going to, you see what I'm saying? Sort of a knowledge pool that eventually would peak. Right. And so the second thing that we haven't quite seen yet, and this is just starting, is what happens to the advertising model. And I know I'm sort of way off script here from, from recruiting, but like in an AI world, all the Internet today is powered by ad revenue, right? In some flavor. Aaron Elder [00:39:29]: Makes sense. Right now your agent or your LLM will very definitively tell you the wrong answer. Six months from now, that wrong answer might be informed by an ad that paid it to do. So. Ooh, makes sense. So it's like, yeah, you're be like, hey, I'm looking for a really good car. Well, surprisingly, Toyota is the first thing it recommends and it's very detailed about its recommendation and why it's better than the Honda. Kortney Harmon [00:39:54]: Because they might have paid more money. Aaron Elder [00:39:56]: They might have paid a little more money to OpenAI to sort of nip that in there. Right. And, and I don't know if they're going to present it to you because as long as it's all factually true, if you will, who's to say it's an ad? Right? It's just so this is, I think it was. Kortney Harmon [00:40:12]: That's a really interesting thing to think about. Aaron Elder [00:40:14]: You can see people sort of trying and wrestling with it. It's going to happen. It's going to happen and that's going to be a whole new weird thing to deal with. Should be fun. Kortney Harmon [00:40:24]: Wow. Okay, now you just got my brain hurting for a 4 o' clock on a, on a Tuesday. I do have one more question. Aaron Elder [00:40:30]: Okay. Kortney Harmon [00:40:31]: And it's a two part question. So if you could send a message back to recruiting business leaders five years ago, knowing what you know now about this transition, what would you tell them to prepare for? Aaron Elder [00:40:42]: Okay, if I can go back five years, Holy cow. Mantine has flown. Here's the weird thing. If you say get ready for AI, they'll be like, what's AI? Kortney Harmon [00:40:50]: Correct. Aaron Elder [00:40:51]: It wouldn't even make any sense. Right. Kortney Harmon [00:40:53]: Okay. Do you want to say two years to make it more like, I don't know, approachable? Aaron Elder [00:40:57]: I don't know, I'm curious what you would say. It's so funny. Like, I mean, obviously go buy Nvidia stock, right? I mean, that's the first thing you got to go tell yourself. Kortney Harmon [00:41:07]: No, no, no, I'm saying in the recruiting industry, silly. Aaron Elder [00:41:11]: Yeah, yeah, yeah. No, I don't know. I mean, as it pertains to AI Because I don't know if you could actually have gotten ahead of the curve in any meaningful way. Right. Is the problem. I mean, hundreds of billions of dollars are thrown at this problem, and if you were trying, you're just gonna get trampled. Right. So things just got really expensive real quick. Aaron Elder [00:41:29]: I mean, if anything, you know, laying low and then, you know, being our first mover on the other side might have been the best play. Right. If you think about it. Yeah. I don't know if. Did you do anything, anything with it? Smart. That comes to your mind on that one? Kortney Harmon [00:41:41]: I think it's. I'm with you. It's like, be the first to act when something like that, like, don't just be jaded to say, oh, I don't know if it's going to work, or, oh, I don't know, adapt. Figure out what's going on and kind of keep moving with the times, whether it's right or wrong, because you're going to be able to adapt easier, I think, in. In that process. Aaron Elder [00:41:59]: Yeah, well. And I mean, even still, we're adding to that. Like, if two years ago you went all in on a. On a V1 of some of these agents, you were going to be pretty disappointed and spend a lot of money on it. Right? So, like, there's a cost of being too early, and that's even today. Right. Like, I mean, keep the task narrow. What was part two? Kortney Harmon [00:42:17]: Part two is what is your advice for the next two years for these same leaders to pay attention to? Aaron Elder [00:42:24]: Definitely hop on the train with appropriate, cautious optimism. Right. Like, I know if you're not using some sort of AI in your daily life, I feel like you're being left. You're just not going to keep up, period. Right. So I think my advice would be think beyond, though, just the KPI wheel and don't confuse change of progress. That would be my advice. Are we just shifting the KPI wheel to have the word AI in front of it? Now it's the AI KPI. Aaron Elder [00:42:56]: There you go. Coin a new term for you. I wouldn't focus too much on that. I would focus on where do you want to be two to three years from now as a company? How are you going to use AI to help you get there, but focus on selling those problems that you see in your industry that are two years out, things are changing fast. You gotta be thinking about it. Now would be my thought. Kortney Harmon [00:43:16]: And do you think you need to have a living platform to execute that properly? Aaron Elder [00:43:19]: Of course you do. Of course. Yeah, of course you do. Kortney Harmon [00:43:23]: Sorry, I couldn't help it. Aaron Elder [00:43:25]: Well, no, I mean, it's funny. I mean, speaking of that, I mentioned this earlier. Like, all of the conversations you've had paired with your ability to go have new conversations is your only real secret sauce. Like, everyone has the same tools and the tools are getting better and easier to get to all the time. And so, you know, our dreamboat, the living platform is to really help you unlock the goal that is your only real proprietary asset and to let you deploy agents internally on your data to do it. I mean, I mean, that's the. That's the whole idea there. Kortney Harmon [00:43:53]: I love it. Aaron, I don't have anything else to ask you. I have kept you for dang near an hour and I have picked your brain and probably stretched it more than it needs to this late on an afternoon. Thank you so much for your time. I greatly appreciate it. Key Takeaways it's not about preparing for change. It's about adapting to change and what's already reshaping our industry. So hopefully our listeners, I'm sure they got great pieces of information and things to think about from our conversation today. Kortney Harmon [00:44:20]: So for our listeners to fde, thank you so much for listening. Let us know what you're hearing and seeing in the industry. And until next time, have a day. Great guy. I'm Kortney Harmon with crelate. Thanks for joining us for this episode of Industry Spotlight, a new series from the Full Desk Experience. New episodes will be dropping monthly. Be sure you're subscribed to our podcast so you can catch the next Industry Spotlight episode and all episodes of the Full Desk Experience here or wherever you listen. Aaron Elder [00:44:54]: Sam.