Episode 132

Max Armbruster on Embracing AI leads to Managing Hiring Mistakes and Evolution of Recruitment

TIM: We are live on the Objective Hiring Show today. We're joined by Max all the way from Dublin, Max. Thank you so much for joining us.

MAX: It's a pleasure to be here as objectively as I can be.

TIM: I love that. And Max, you run a business called Talk Push. And I'd love to hear just a little bit about yourself and the business just so our audience can start to think about who they're listening to today.

MAX: Yeah, I'm the founder and CEO of TalkPush. It's a recruitment automation platform. We're 10 years old, and for the first eight, we called ourselves a recruitment automation platform. For the last two, we did what everybody else is doing, and we're calling ourselves an AI platform. And we were waiting for the moment to be able to call ourselves that, but the technology made it possible a couple of years ago. And now what do we do? We automate about 10 million job applications a year through the conversational AI technology. And as our customers have come to rely on our technology to engage with candidates, it's given us an option—not really an option, a mandate—to go from being a prescreening tool to being a sourcing tool, a prescreening tool still, an assessment tool, and an onboarding tool, and once you have a communication channel open with a candidate, that communication channel can be used for a lot of things. So we've expanded into the candidate journey to be more end-to-end.

TIM: That's exciting. And so you're the founder, and you've been running this business for more than a decade now. How's the journey been?

MAX: I'd never held a job for more than five years before that, no, less, four years. That was I never thought I could hang on to a job that long, but it has kept me on my toes. It's definitely been, I, people ask me two things that people comment about entrepreneurial experience. They, a lot of people will say, Oh, I really want to do that. I want to start my own company. My go-to answer is don't do it. And then the other thing I hear often is, You're so brave. You're such a risk taker for starting a business. And I say, no, you've got it wrong. I want it to be in a company where. Where the boss couldn't fire me, so I want to be in control. I don't want risk. I want stability in my life, so Yeah, two common misconceptions about entrepreneurship, at least as I experienced them, but yeah, the journey has been a bumpy ride, and you know, the only way you learn so much is by making mistakes. So, it's rewarding in that sense.

TIM: Yep. Yes. And I'm sure we've had plenty of shared experiences in the way you just described. Then it really resonated with me. And this is also the longest I've ever held onto a job easily. Six years now for us, almost to the day; it was our birthday a few weeks ago. Congrats. Happy birthday. Also, I enjoy the control and being somewhat in charge of my own destiny. It's at some level, or at least the feeling of that.

MAX: Yeah, it's It's actually good to be, if you're a bit of a control freak, it's actually good to be the owner. And I didn't know I was that; I saw myself as a highly creative person. And of course you have to be creative too, to be an entrepreneur as well. But I think it's that balance, finding a job where you could be both creative and in control and working with others and working in tech. Those are a lot of Ns, so it ends up, there's just not that many jobs where I could be happy, but I'm very happy in this one.

TIM: One thing I have always found interesting about entrepreneurship is. Just the breadth of different things you get to do if you're the founder, it's almost unlimited in a sense, and then also having to get quite deep in several areas. So it's a weird combination of a lot of breadth. And then in some cases, a lot of depth as well, especially in the early years. I would have thought, did you find that?

MAX: Yeah, I think it plays well for people with attention deficit disorder who need to move from task to task. You'd recommend for most professionals to lock out a big section of their schedule to do deep thinking, spend time on things, and resolve big tasks. But if you prefer to be somewhat reactive, a bouncing electron CEO might be the job for you. Because you know, it's your kind of surface-level understanding of things, but you have to move fast. I don't, by the way, I'm not going to write any management books on how to be a great leader or a great CEO or anything like that. I have nothing to, I have nothing to, no wisdom to impart into the world on other than my subjective experience of it. Take all that with a grain of salt, but I do think. Being able to bounce around fast is a good skill to have for this job.

TIM: Yes. And again, I just feel like we must've had so many shared experiences. I literally did it about one hour ago. Bought a book called ADHD 2.0 from Amazon. And it's going to help me with some of those focus issues. I feel like, actually. Running a company has maybe made me more ADHD, if just because there's so many things happening all the time. I don't think I used to be like this. I used to be able to really focus for long periods of time. I remember even at university, sitting down there for five hours, doing matrix algebra and calculus, undistracted. So I feel like there's some combination of running a company, technology, and dah that maybe makes it feel like. You were distracted easily or something.

MAX: Yeah. Maybe take off the notifications on Slack and email and stuff like that. But I agree with you. It does. I think the whole world is moving in that direction. So hard to say whether we're far from the baseline or not, but it does feel like we're moving in that direction mentally.

TIM: You mentioned at the top of the call learning from mistakes in business, and one big bit of business story, of course, is hiring and getting the right people. And I'm sure anyone would say. The difference between a great hire and an okay hire is astronomical, at least in my experience. And I'm sure we've all had people that we hired that we realized we made a mistake. I've certainly been reflecting on this myself recently, and it's something that's come up on the show a few times. How about for yourself? Is there a particular hiring mistake? Of course, not mentioning any names that you've had in your career that changed the way you hired.

MAX: too many to recall It's a kind of a blur at this point, but generally, my heuristic coming out of these 10 years is when you're a startup, hire people who? are multitaskers; hire them because they're in love with your product and do not try to bring in management too soon. It's generally much more advisable, especially if you're bootstrapped and you don't have other people's money to spend to grow your management team from within and to promote from within and to give somebody who joins you in their twenties an opportunity to manage in their thirties, something like that, rather than parachuting managers from outside. I think that's generally good advice. I would go there, and then, yeah, a lot of the money we spent on was bringing in executives who then had to spend the next six months of their lives trying to figure out how I work and how other people work. You don't have that kind of time scale in six months in a startup life; it's just so long that you're really starting from, you're starting these; it's not an ideal start if it takes six months for you to understand how everything works. And unfortunately that's often the case in this environment. So that's it. I could go into, yeah, I could answer this question literally in a hundred ways. So many mistakes. name?

TIM: Hiring for a startup is hard; hiring for your own company is hard as well, where it's your own money. I'm interested in that lens; actually, I'm not sure if you can remember back when you were doing a regular job. if your philosophy on hiring was different compared to being the owner of the company where it's your money at stake, and it's your name on the side of the building, so to speak.

MAX: When it's your name on the building, you want to work with people that you like, and that you would enjoy spending time with, and that you feel like they sound a little bit like you. They have the same level of value sets and the same level of intelligence. And whereas you might be somehow more task-oriented when you're hiring as part of a corporation, meaning skills-oriented in a way you might be more objective. If you're hiring for someone else because there's less of an emotional attachment, but yeah, it was so long ago. I really can't I really can't recall. I just, all I remember from back when I was hiring for others was that it was very tedious and boring, and I hated it. And that's one of the reasons why I started a recruitment automation company: because I thought it was so boring and repetitive. And recently someone asked me, What's your favorite part about your job? and going full circle. I told them my favorite part of the job now is hiring, is recruiting. So it's ironic that I hated recruiting, so I automated it, and now my favorite part of the job is actually recruiting. So full circle.

TIM: And is that because you've automated away all the clunky, annoying bullshit stuff?

MAX: somewhat. I mean, it's nice that I'm able to listen to candidates and filter the bad ones. But it's also because I guess I've made, going back to your previous question, I've made so many hiring mistakes that now I can internalize every time I find a bad candidate or a candidate that's not a good fit. I get a little jolt of happiness from avoiding that mistake. Be like, I recognize that mistake. And if it gives me a bit of adrenaline to be like, Thank God I was able to spot that," it's like the mindset of a QA person where they get joy from seeing a bug. I get some sort of a sick pleasure from saying no to somebody.

TIM: I know what you mean because I think my, my, probably the best skill that I have as a founder, at least in a product company, is a real sense for identifying product bugs, especially in the first few years, identifying and predicting. I bet you we would have forgotten this particular corner case. I'll test this bang broken. I feel like I really got a heuristic over that. Maybe not as strong of a people, but. Because the sample size is smaller, like we've had, 8000 tickets in Europe, but we haven't hired 8000 people. But yeah, I've had my fair share of mistakes as well. I wouldn't mind sharing one briefly, actually.

MAX: Yes, please.

TIM: The first person I ever hired, which was now 11 years ago, so in a company I was working at, I was trying to hire an analyst. I was working as a commercial analyst. I wanted to hire an analyst to come in and do actually something similar to you in a sense, do some of the boring bits of my job, which was tedious manual reporting. We had this ridiculous scenario every month where we had to combine eight different spreadsheets into some kind of super report spreadsheet for a CEO who wanted his numbers. And this process took days. I wanted to hire someone to take care of that and other things. And I remember interviewing people and them the role and saying, Oh, like, we've got a lot of interesting data here. That's untapped. Like we haven't looked at X, Y, and Z. There are all these interesting bits of analysis we could do, which was true, but I didn't say to them, Oh, by the way, to be fair, the first four months are going to be extreme tedium. And your job is to automate away this garbage. So then you can do more interesting things. So I over-pitched the role. So the first person who came in quit on day five. Like I'm not going to, I'm not a spreadsheet monkey. I'm not going to do this if I'm sure of what they're thinking. Now I'm not the fastest learner because I went straight back to the market and went through the whole hiring cycle again. Hide someone again; they left on day two. I'm not joking. Okay. I

MAX: Now you remember 11 years later.

TIM: Exactly. I remember my boss calling me and saying, Oh, the worst possible thing has happened. I'm like, Oh, watch that. So-and-so has quit. And okay, great. And then I was demoralized because, of course, this is months of effort, and I'm now back doing the thing I wanted to automate away.

MAX: Tim Why did this person leave after two days? What did you do to this poor person?

TIM: I swear to God, it was the job itself. The fact that I made them sit there and your copy-and-paste shit in Excel. I wasn't actually a psychopath boss, but you can have your own theory. I remember sitting there demoralized at my computer and hearing a young guy behind me who'd been hired just for the day to do some temporary filing work with the HR team who was next to us. And all I heard him say was something, Morgan Stanley internship something. And so my brain was like, Oh, okay. He's probably smart. That was my heuristic. Like, at least he's probably clever. If he's just done an internship in mergers and acquisitions at Morgan Stanley in London, he's probably not a dummy, which I think is a reasonable assumption. Anyway, so I pulled him into an interview room and just told him about the job, hiring him on the spot. He lasted like 10 years in this company and went through the

MAX: there you go. Phew.

TIM: leading the data function. So even the dumbest hiring process sometimes can work.

MAX: Yeah. And go find out those Morgan Stanley interns. Yeah. Did he, or was he pre- or post-intern?

TIM: Yeah. He did his internship at Morgan Stanley, and I hired him as basically a full-time analyst.

MAX: Okay. So he'd gone through it. Okay.

TIM: Yeah.

MAX: It'd be even better if you could just identify who gets shortlisted for the internship. And you don't have to actually do the internship. They did all the hiring for you. But okay.

TIM: Now. Thankfully, in most cases hiring has become a little bit smarter since then in the last 10 years; at least the opportunity has to make it smarter. I'd love to get your general thoughts. On the state of AI and hiring, like, where are we at? What can a large language model do in hiring? What can I not do? What's your kind of lay of the land overall?

MAX: I'm very bullish on what he can do. And we released products that can conduct interviews that I would say are of a very high caliber, meaning I would feel generally better about this AI interviewing a candidate than asking somebody on my team or even asking myself. Yeah, I would, I definitely would. This AI can interview candidates better and more consistently than I can. And I think it's yeah. I think the technology, we're in one of those moments where the technology is ahead of the expectations and where people just haven't absorbed the fact that they can do Every round of interviews, from psychometric assessments to culture fit to technical assessments, they can ask an AI to do it. Very reliably and obviously at scale very cheaply. And it seems that they haven't accepted that, obviously we're technologists, so we're always happy to try new things and push the boundaries, but I guess HR is going to be one of those areas where it might be lagging; it's going to take over sales and marketing before it takes over HR and talent acquisition, which buys us time to build a wonderful product. So, so, that's nice. I think your question was a little… you were expecting a more technical answer, and maybe we'll get there, but just setting the scene for You know what we've learned over the last few months.

TIM: Yes. Yes. And I'm interested to hear why you would imagine it would be adopted earlier in sales than HR. Is it because sales is a bit more black and white and brutal? It's this tool that makes us X dollars. Of course we're going to use it. Is it, is that part of the difference in the almost like the culture between those teams?

MAX: Yeah, they do run sales, run more experiments, and have more budgets. And because they affect the top line of a business, you can always run an experiment and say, If this doesn't work, you know, this could make us a million dollars, and it costs 20,000 to run the experiment, so let's just roll the dice because the expected outcome is positive. You don't have that mentality in recruitment; nobody wants to roll the dice because, yeah, nobody wants to roll the dice in that team. No risk takers. With exceptions, of course, and special kudos to all of our customers who signed up for TalkPush in the last 10 years, who did take a certain degree of risk by choosing a relatively young provider.

TIM: I wonder also about my observation. Also, working with talent teams over the years has been okay, so maybe not particularly risk-loving. It's fair to say, but also in some sense, a lack of diversity of mindset and background skill set. So you know how many talent leaders come from, I don't know, a hard science background or data analytics or B.I. or software engineering. Or something like that, I feel like, would then give them a slightly more technology-datary AI view of things. Whereas, yeah, most of them wouldn't be from that background. Is that fair as well, do you think?

MAX: Yeah, generally, more soft, like social science and

TIM: one

MAX: With primary concerns about how it will impact the mental state of the candidate and convey a strong sense of human connections when it comes to the job, absolutely. There, there's another dimension to recruitment, which is if you don't go to a top university, but you're a hardworking hustler. It's one of the best jobs available in the industry because you could really just work your way up, and it's a high volume, it's a numbers game, recruitment, just like sales. And the idea that that young, that job, which is perfect for an early-career ambitious person, will be automated away. is is, depressing for a lot of people who've since reached management levels and talent acquisition because they're like, where would I have begun my career? If I couldn't do these interviews, if I couldn't, if I could source candidates on LinkedIn or job boards, where would I have begun? And obviously we don't know the answer because it's science fiction stuff, but that's another reason why there's some pushback on this technology.

TIM: comment I've often heard as well is this sense that in recruitment, adding technology is somehow dehumanizing; for me, if it's done right, and it's good technology, it's the polar opposite, because what could be more dehumanizing than a hiring process completely devoid of any automation, where everything falls through the cracks, and of course people forget to follow up, and blah, blah, blah, and this is why we get into a scenario where candidates get ghosted all the time, and that's their number one complaint. Like, it's because that's such an easily solvable problem. So I personally would like to push back on that assumption that technology dehumanizes; I assume you would also agree that it could actually, humanize the process in some sense.

MAX: Yeah, I could. I think I want to take a different angle and say that you want to have a human experience. Humanizing, whatever that means, is a humanizing experience, a high-touch, warm experience that resembles the one you would have in a small tribe or in a group of a gathering of people. That's supposed to happen at work when dealing with candidates. I don't know what your candidates-to-hire ratio is, but I'm hearing across the board that. That ratio is increasing, whether you're talking about general leads or qualified, hyper-qualified profiles. The numbers are increasing because it's never been easier to apply for a lot of jobs. And now there are AI bots that can apply for you and write your cover letter and write your resume for you. So you're going to pass all those. All those screenings automatically, very easily. And so the number of candidates is just going to go up and up and up. In that context, there's simply no, there's no other way. Either you use technology to overcome this increase in volume, or you're going to get left behind. You're going to get the last remaining candidates, the candidates that nobody wants because your recruiting time will be weeks or months instead of days. The debate is about what, obviously, we should make something that feels good for the candidate, and there are ways to do that with technology. I don't hear anybody complaining about their dehumanizing experience on Uber. Oh man, I wish I could talk to someone because it's smooth.

TIM: Yes,

MAX: If we make it smooth, that's good enough. And then, if you don't want to use technology because it's dehumanizing, that's okay, but you're going to be out of a job. pretty soon.

TIM: harsh and fair. And I wonder also if one element to this is that candidates, because they're individuals, were so quick to adopt AI as soon as ChatGPT came out; I'm sure candidates were using it to optimize their resumes and what have you. As you say, now there are tools to apply automatically as well. And so companies are just lagging behind candidates by a few years, in a sense. With the fact they haven't necessarily adopted the same level of AI in

MAX: Definitely in the talent pools that you navigate, Tim, I would expect, if you don't have a perfect resume and cover letter, then what do you get? What are you doing in this space and in data and business analysts? You shouldn't be working here. But maybe in some other categories, for example, in retail. Some employers are not lagging. It's the other way around. They're ahead, but the candidates may be lagging. But yeah, some candidates, it's very tight. Actually, I think there's just a lot of range for candidates. You have some very smart, very good candidates that have never been on ChatGPT, and there's tons of them. And then there are some people who wouldn't be able to spell their name who are experts at it. That really creates a lot of variance. I

TIM: And that's a good point, though, because, yeah, if you're in this sort of startup data tech bubble or vacuum or whatever, you would get the idea that everyone uses ChatGPT all the time for everything. But clearly that's not the case.

MAX: mean look at the number of active users. It's very impressive. I haven't checked the latest data, but we're talking about hundreds of millions, not,

TIM: Yes.

MAX: Not billions, Google.

TIM: Yes. What about in terms of interviewing specifically? What are you currently seeing? AI as an interviewer versus, let's say, a more traditional human interview.

MAX: Advantages of the AI interviewer are that it is able to work faster with all the context of all the conversations, everything, and all the interactions that you, as an employer, and the candidates have had are in its awareness. So it will never ask a candidate to repeat itself unless you specifically design the AI to ask the candidate to repeat itself. So it has a level of awareness of the candidate, which, as a human, you don't have. You may have read their LinkedIn profile quickly before the interview. You may have browsed through some interview notes. You may have even looked at the communication history in your ATS. But not to the degree where you know everything, every word that's been said by the candidate from the moment they heard about you to the moment they're talking to you right now, this minute. So that level of awareness is much higher. Obviously, AI has also read the equivalent of 3,000 lifetimes of books, like a model like ChatGPT, which has read the equivalent of 4,000 lifetimes of books, so it can interview you for any job under the sun, say interview me for a nuclear physicist position or for any position. So it's just got immense range, whereas recruiters have to do a really deep dive into a domain in order to sound halfway competent. Even somebody who's been recruiting for IT for three or four years, if they're not an engineer, the candidates are going to smell it within 10 minutes.

TIM: Yes.

MAX: Yeah, and there are other dimensions, but I'll stay there for, yeah, so I don't take the next 20 minutes reciting them, but yeah, these are two new ones.

TIM: Yes. And is part of the issue then that I'm just trying to think why, like, why haven't we necessarily accepted where we're at with AI? Is it because it's happened so quickly? Is it because the implications of accepting how much it is and how good it is are too profound for us to even contemplate? Is that part of the problem?

MAX: I think so. I think it's when we discovered that the Earth was not in the center of the universe, and then we discovered that the sun was not in the center of the universe. And these discoveries were made, and then they, it took hundreds of years for humanity to absorb these shifting moments where, and I think we're in the middle of a shift where, yeah, we're not the smartest entity anymore in some domains. We may still have the lead in a few domains, but we definitely lost a chess, and I think we're about to lose that interview. Or in the midst of losing an interview, and these are not easy things to recognize and accept. So I think that it's just going to take maybe a generation, 20 years, and by then, of course, the technology would be so good that it would be undeniable.

TIM: I wonder if it's also just once we're interacting. our day-to-day lives more with this technology, it will be completely normalized so quickly that then the idea of, Oh, I'm getting interviewed by an AI rather than a human, will be just like, Yeah, of course, like why wouldn't I? I interact with this every day in other domains. Maybe that's part of it.

MAX: Yeah, it is a con. We have to look at candidates as consumers first. The way they consume information is determined by the broader behavior as a consumer, the way they interact with things, and the career websites and the job application form. The interview, all of these things are downstream. First you look at how consumers behave, and then you adapt to them. And you can't dictate; you can't tell consumers how to behave. You just have to look and observe and wait and see. Yeah. But it's going to happen pretty quickly. I've, I don't know if you have kids, Tim? No?

TIM: But one day.

MAX: I have two little kids, one three-year-old, and I can put her to have a chat with ChatGPT on a topic, and I buy myself 30 seconds where I could go and do some, run some errand or something. And she's completely consumed by the experience. It's very easy. They're growing up with that stuff now.

TIM: Yes. So there'll be AI natives effectively, and it'll be normal for them. I'm wondering if you could shed a light on this, actually. Who is more reticent around AI interviewing? I'm assuming it's the companies having, like, an overconfidence in their superior ability to interview, as opposed to the candidates experiencing going, Oh, I don't like this.

MAX: Yeah.

TIM: Of the two different sides?

MAX: Yeah, I'd say most of the pushback is on, yeah, it's almost, I was going to say, I agree with your premise, but there is still quite a lot of work to do on optimizing the experience for the candidate and eliminating all the edge cases for an AI to interview in the most natural way possible. And small little things get in the way. If you operate in a market that is bilingual, most voice AIs are not so fluent at moving from one voice to another. If you're talking to a candidate in a noisy environment, the AI gets curved. So these are technical issues that are trivial when you're experimenting with this technology in a lab but become very problematic in the real world. So yeah, there is pushback in that sense where some candidates you just have to iron these things out. So I'm glad we have some time and the candidates aren't, that companies aren't all begging for this thing to go live at scale. But then, of course, mostly it's they try; buyers try it, and they like it. And then. And then they're just a bit overwhelmed by the implications because it means massive restructuring takes, and you have to plan it the year ahead. So it takes at least 12 months to restructure a company. And so considering this technology is nine months old, we can't expect amazing results, right? We have to wait for companies to rethink how they're going to organize themselves. It takes years. It takes years for true automation to take place.

TIM: And so that's just a straight-up part of the discussion: this technology is directly required for a restructuring because it's going to automate away these various things that are just inherent.

MAX: Yeah. If an organization does not want to have that conversation, they say, No, we want the AI. We want the cool stuff, but we don't want to talk about restructuring and reorganizing and moving people. Then I'm like, okay, I don't know how to, I don't know how to help you. So yeah, go play in your own sandbox because we want to have real impact. If there's not a real business driver for it, then it's an initiative that probably has only a few months of existence before it gets replaced by something useful.

TIM: So it's almost like one of your qualification questions in a way.

MAX: Yeah, it sounds brutal, but yeah, we could say, Okay, but increasingly I think the market has shifted in 2024. There's been a lot of press around restructuring and doing more with less. And I think executives are more comfortable having that discussion up front. Compared to a couple of years ago, a couple of years ago. We like, whatever you don't mention it, it's human nature. You spend a lot of time hiring people. The last thing you want to talk about is letting go of half of them. It's not something that you enjoy talking about in a public forum with a potential vendor. It's not a, it's not a nice conversation to have.

TIM: You mentioned voice AI, and I myself have been using Chattopadhyay a lot recently for learning languages, and it's pretty amazing as a free language tutor. I got to say, like on the voice mode, switching between a bit of Spanish, a bit of Italian, and a bit of English gives me a quick quiz. Translate from this to this, even like mnemonics. Oh, here's this list of words. Give me a little clue for each word that I can, and will help me remember the

MAX: Nice.

TIM: It's

MAX: Nice.

TIM: Staggering, and well, it understands me in English, because for years, Siri and those kinds of tools, like, they're still pretty crap, actually. But these new versions are staggeringly accurate from what I can tell. This really is part of the enablement for the interviews.

MAX: What are some what's? What are some exceptions to what you've just said? Some moments where they fell short. Do you recall any?

TIM: Not typically. One I had today was I was trying to, because I read this on LinkedIn, that maybe it would be a good tool to tell you how to pronounce someone's name. So I had someone who I was about to speak to, and I gave it its name, and it was way off. So I'm glad I asked at the start of the call, how do I say your name? And then they told me, so that didn't work so well, but like in terms of the—that's a good one. Really good.

MAX: Yeah, nice. Yeah,

TIM: Picture of a written book, and I like to underline the words on the paper that I don't know, and it can pick up those words and translate them. It can identify which ones I've underlined pretty well. As long as it's zoomed in far enough. So I've, yeah, it's just been remarkable so far.

MAX: The name thing we've run across is the detecting of emotions. It's not easy. So most of the voice models are not trained to recognize if somebody is sobbing or smiling. That's hard to detect even though it's been promised by OpenAI, but I think the promise is not fully met, and then like right now I paused a little bit in my speech, but you could tell that I was not finished speaking, and that's one of the hardest things to achieve is to train a voice AI to know when to let somebody finish their sentences. And on the flip side, if somebody is rambling on way too much and veering off topic. How to stop somebody from speaking in a pleasant way. That's hard too. And that's some of the finer things in interviewing that I think are where the humans still outperform the AI.

TIM: Yes. Yes. But as you said before, your kind of overall view of AI interviewing now is that it's definitely a net win at the moment for the AI. So in terms of, you mentioned like consistency in terms of having that context in terms of being on an interview for any role in the world, as opposed to just like your area of expertise. But maybe not picking up on some of the finer human details. That's the kind of lack at the moment.

MAX: It's still more natural to have a conversation with a human than with an AI for the reasons I mentioned. But the AI will be better because it's going to ask all the questions; you'll never forget anything; you'll record everything. It's just more reliable and, of course, much more scalable. And it's immediate. If, blue pill and red pill, Mr. Candidate. In the red pill, you're going to talk to a human being in the next couple of weeks, and he's going to interview you for 45 minutes on Zoom. Red pill? Wait, that was the red pill. Blue pill is Blue pill is you're going to talk to an AI right now who's going to tell you within 20 minutes if you are a right fit for the job or not. Mr. Candidate, which one do you want to pick? I'd say it wouldn't be 100 percent blue pill right now, but if you were talking to people below the age of 30, you'd probably get 80 percent blue pill, and yeah, maybe if you talk to people over the age of 50, that'd be 80 percent the other way.

TIM: Yes. Yes. It's exciting times. But also the technology hasn't really been rolled out as quickly as what you might've thought in talent, given the state of large language models now, as you say, can solve a lot of the problems. Is there too much regulation, too much red tape in place? Do you think?

MAX: a resounding yes. But this regulation is one that lives often in a different world from reality. It lives in a sort of political world where these laws are passed, but they're not, first of all, they're not written in a way where they can be enforced. And they're mainly a posturing statement to say we are looking after you and to talk about ethics. The ethics of AI making decisions, which I don't know, I think it's a, I think it's a political play. I don't quite understand why asking an AI to review somebody's skills or resume is an ethical question. I think that ultimately we all know that the human will make a hiring decision. And if you buy a tool that makes a predictive assessment of somebody's suitability for a role, you should pick the right tool. That's obvious. And I think that's something that's going to be figured out by the market, which tool is the right tool, because it's going to give a huge advantage to the employer. So why do you need to regulate it? I don't know. I think it's very political. I think that there's also a way for people to deal with the stress, the anxiety that comes with the changes ahead, and to delay them, which is to say these are all these ethical questions related to AI. Is it fair? Is it racist? An AI is a large language model based on a bunch of human beings that predicts what those next words should be. How can that be described as racist? It's a real stretch, but the argument is strong enough. I think a lot of TA professionals are falling in line with it. And maybe because it's a convenient argument that allows them to postpone the inevitable, which is you're going to have to transform your operation to take advantage of technology.

TIM: Yeah, and I think the narrative around the transparency, auditability, and all these kinds of things is fair enough, but we don't have that at the moment for human-based recruitment. Nobody can inquire and ask a company, Hey, that job I applied to three years ago, who rejected me? What was the reason for rejecting me? Show me the evidence like that doesn't exist as I'm aware.

MAX: It's infinitely more transparent and auditable than a human. It's so much more auditable and transparent than a human. But also, like, it's so easy to keep track. It's so easy to meet those requirements, to have a checklist to say I am compliant with the EU AI Act because I store all the information here, and this is the prompt that I use, and just to put it out there. It's not complicated. It doesn't really add any value to anybody because the model and how you optimize them change all the time. So you're constantly changing it and making it better. So what you're going to spend, half your R&D budget, is just updating documents that nobody's going to read, basically. So that's the end result of all this work. And the other end result is, of course, people who really don't want to buy your solution. Can you audit you to the end of times to postpone it? Without coming across as Luddites, so they don't have to, they can talk about AI. People who don't want to automate can talk about AI and regulation. Until the cows come home and that way, and still come off, coming across as well-versed, intelligent, and positioned in the organization to make strategic decisions. So for them, it's a win. But I think it's a waste of time and money to try to regulate something that is so nascent and new and obviously useful. And I also think it's a disservice to candidates. That we're making it harder for companies to adopt automation at scale because it means that many candidates will simply not get heard at all. Or will just be ignored or eliminated for the wrong reasons. Because the employers couldn't use the state-of-the-art technology that's available in the market because it didn't go through a compliance review. This being said, my company is compliant. We store our data locally. We publish our AI standards and ethics. We're, we've, we fall in line. But we managed to do it rather efficiently, ironically; a lot of the stuff that we have to do in order to be compliant with AI regulation we can get done much faster with AI. If you know how to use AI, then the regulation is not too much of a hurdle because you could just use AI to deal with it.

TIM: Yes. As someone who's filled out plenty of information security forms over the past six years, yes, the godsend of AI doing that on your behalf is thank you. Fighting bureaucracy with AI. That's a great use of AI for me.

MAX: Yeah, so something that is pre-AI is infosec, infosecurity, and we, you know, we work with some of the largest employers in the world, like Accenture and Walmart. So we always go through InfoSec reviews, and it consumes a lot of time from our engineer, or at least it used to from our head of InfoSec. Now we've trained an AI to handle all of these queries, and so our sales team can just ask the AI, Oh, what about this? What about that? And Phil forms automatically.

TIM: Oh, amazing. That is a great use. And yeah, God, I wish I had that five years ago. What is more soul-destroying than filling in a questionnaire about security or this or that? That's just, that's not the, not what we signed up for, is it? But that's what you have to do as an entrepreneur, a bit of everything.

MAX: Yeah. Yeah. You have to find the compliance officer within you.

TIM: Exactly. Max, if you could ask our next guest any question about hiring, what would you ask them?

MAX: I always like to ask people about the hiring mistakes that they've made, which is a question you asked me today already, so I don't want to strip it away from you. So you go and ask that question again, but yeah, other than that, how are they going to hire differently in 2025 versus 2024? How are they going to adapt their hiring practice to deal with the fact that candidates are GPT-enabled?

TIM: A great question that I will level at our next guest at some point next week. Max, it's been a great conversation. I've really loved it. And you've given such a different perspective to every guest we've had. And so it's great to have you on with your access and knowledge. You've got primary data on what's happening right now in the market. So thank you so much for sharing that with our audience today.

MAX: Oh, it was very liberating. Thank you, Tim, and I hope I don't upset anybody.