Jeremy Lyons, Founder at RecOps Collective, Why Your AI Strategy Needs an Ethics Layer Before Anything Else

[00:00:06] Hey everybody, I'm Tony Buffum, Co-Founder of Human Cloud and she's Strategy Officer of Human Cloud and I am here today talking with Jeremy Lyons. Jeremy, we connected through what was PeopleGPT with the common, you know, love and energy around the future of recruiting and the future of HR by, you know, just exploring how we can leverage GPT prompts, that type of language and technology.

[00:00:36] Tell us a little about yourself. Yeah, so Jeremy Lyons, I've been in the recruiting operations space for nine years active. I've been in the recruiting space for a lot longer than that. I am second generation recruiting. My mother was an executive recruiter.

[00:00:52] So, very deeply familiar with a lot of different elements of the talent acquisition space. And over the last couple of years, I started RecOps Collective at the end of 2022 with a very explicit goal of actually helping companies be more operationally sound, more operationally focused.

[00:01:13] So that when AI started coming in, people were able to have a good enough foundation, good enough data to be able to start implementing AI pieces rather quickly. And obviously, that has evolved as everything has evolved over the last couple of years.

[00:01:38] Yeah, rapidly is how it's evolved. That's what I'd say. I think you have to start from a foundation of standardization and, you know, effective, efficient workflows for sure. Tell me a little bit about how you help companies do this and where does all your RecOps mastery really come from and how does it get applied? You know, it's funny because I think there's two elements to that. And I think one of the first ones is that I think a lot of my RecOps mastery comes from pain.

[00:02:09] Because I think that when you're thinking about process thinking and you're thinking about human-centered design and you're thinking about first principles, it becomes really, really easy to say, okay, me, the candidate, what do I experience? Me, the interviewer, what do I experience? Me, the administrator, me, the sourcer, me, the this.

[00:02:33] And you have to have had at least a little bit of experience beyond the theoretical to be able to say confidently or design confidently a process that allows for that standardization. Now, the other thing too that I would say about how do I help companies do this? Well, I think that one of the very first things that I ask a company whenever I'm talking to them and they're talking about,

[00:02:58] hey, we want to use AI or we're thinking about implementing more sophisticated solutions is I ask them very clearly to articulate what is their AI strategy. And one of the reasons why I ask it in that way is that there are functionally two choices when you think of your AI strategy. There is, you're going to design your AI strategy to be a co-pilot. It's going to be there. It's going to be there to help your teams.

[00:03:24] It's going to be focused on how do we not only elevate the user or like the end user, but how do we evolve it kind of like symbiotically? Um, that's why I think when you look at something like a Zapier who has been very public about like, this is our AI literacy. This is how we are doing it. And this is, this is how we think about it. It's a very, very great foundational doc.

[00:03:50] Arguably what I would say is, you know, people talk about, oh, Netflix's culture deck. Well, I think that the, as being this like transformational document in all of Silicon Valley, I really think that what Zapier has stated, how they have stated it has even more power than that. But on the flip side of that, you have to acknowledge that some companies are not building it to not using AI to be a co-pilot.

[00:04:19] That they are genuinely using it to, I am going to replace something, whether that is the humans, whether that's an entire system. And you need to be able to be flexible to do both. And I think that when certain people, I would say, might view that that output is really, really dirty and say like, hey, look, like, I don't want to be a part of this. I don't want to be, it's sort of like when you recruit people into defense companies.

[00:04:47] And they say like, look, I don't want to, like, it's a great opportunity, meets all my requirements, but I don't want to be in charge of a missile guidance system that might be, you know, I couldn't sleep with myself at night if I did that. And that's, again, that's okay. I'm not saying one thing is right and one thing is wrong. You just have to be okay when you're having those conversations up front to be able to say like,

[00:05:12] actually, you know what, no, I don't agree with a co-pilot solution. I'm all in on this is going to replace stuff and I'm going to build you stuff that's going to replace stuff. Or you're saying, you know what, building like actually genuinely replacing human beings is not in my interest. I have no, I don't want to do this. That's not aligned. And I'm going to do that. And I think the biggest thing that I see whenever I have conversations with people where they're not sure is they're sort of like, well, show me what good looks like.

[00:05:42] And you're like, good is going to look very differently in these two situations because not only are we foundationally having to rethink our system and how people think, we have to do so much more than that. Yeah, I think it's a really good point. There's a pretty broad scale in terms of how companies think about their AI strategy and where in their organization they prioritize getting implemented. I was telling you earlier, I just got back from Shrong Talent.

[00:06:12] And at that event, of course, AI was the forefront. I looked at, I did analysis of all their session listings. Over 70% mentioned AI in the title or description. Not really surprising. But I think what is surprising is there's a lot of theory and strategy, but not a whole lot of real world application. How do I use this every day? Which is something I think you help folks with.

[00:06:39] What do you feel like is the gap between that strategy and the actual everyday application? So I think the gap, so I think that there are two things at play, maybe three. I think part of the reason why you see it on all the conference schedules is that people know that people want to hear about AI. And so it's like, hey, we have to have something with AI here. And I help companies plan conferences all the time with rec ops people involved.

[00:07:08] And, you know, it's like, well, no one will show up if we don't mention AI. So we need to talk about it. And it needs to work itself in there. And so that is, I think, there's the marketing standpoint of that. And I think then the other part of it becomes there's an education standpoint, which is to a lot of people, especially in our space, this is totally new, new ground.

[00:07:31] And so you have a lot of people posturing that they are true experts in what they're talking about, just to get on stage to say that so that they can position themselves. And that is, you know, something of a disservice there. Now, difference between the original question, which is the how do you take it from theory and speculation into genuine action?

[00:07:58] I think that there are one of the, a couple of things that have to happen there is you have to take a position of encouraging curiosity on your team. So, yeah, and I am a firm believer of you can create the right environment, but the right people who are curious, that that is a deeply intrinsically motivated thing. They have to be able to say to themselves, hey, I understand myself well enough to be able to start trying to explore.

[00:08:28] But then the other piece on that, too, like the people who are, I think we're seeing this, there was this old model that used to exist where you had, you got into a leadership position. And as you sort of moved out of tactical to strategic, you didn't get your hands dirty anymore. And so you kind of sort of got into a place where, hey, okay, we're going to launch a referral program,

[00:08:57] or hey, we're going to change this thing. And then you delegated that. And I think that right now, people are going and saying we need to have some senior managers or above talking about this thing. But the people who are really sort of figuring out the boots on the ground stuff are the tactical people from like the coordinator or the this or that. And that part is hard because it's this shift.

[00:09:27] It's this power shift. And people, I think, are still adjusting to what is the power shift between the strategic output with the actual tactical output that you're going to need. And you see the prompting differences between the groups. And that's something that I also pay attention to, that your coordinators, your specialists are prompting or thinking very differently about it

[00:09:50] than your people who have 10 years of experience who might have difficulty articulating how they developed a complete thought that once was a tactical thought on how to do it. When you think about that tactical application, what are some ways that, whether it's augmented by a co-pilot or just using AI and automation to reach a broader group of candidates or provide a more customized candidate experience,

[00:10:19] what are some ways that you see an effective application of AI in recruiting right now? Yeah, no, and I think that that's a fantastic question. And I think kind of to build on my response to the last one, I think that what we see is a lot of people have kind of determined that a good jumping off point for experimentation is job descriptions, candidate reach out,

[00:10:46] and maybe a little bit of market analysis, data analysis, because that's sort of the start of the process. And I think that it depends on how you think about a problem, because what I kind of see a lot of people do, and it gets to the tactical elements of how do you do it is where do you start from your solution? I think there are a lot of people who take, I'll call it like a post-mortem response.

[00:11:15] We've gone through this a couple of times. These things broke. We know this because of the end. We got to the end. We're exhausted from the marathon. Now let's go back and see like, what did we do at mile marker 13? What did we do at 20? What did we do in the last 286 yards? Now let's go and fix that.

[00:11:36] Then there are some people who stand at the front of the process and they go, okay, at those same bullet, at those same points, because we've done it, these things are going to happen. And let's go and solve those solutions now. And that involves a certain level of a company. And, you know, eventually you kind of call it, well, that's foresight. And oh my God, this person thought of this something ahead. It's like, yeah, because this person has been in this thing so many times.

[00:12:00] And then you get kind of a middle out solution where they kind of say like, hey, there's something wrong in our interview process at the hiring manager stage. How do we solve for this? And one side goes running towards the end from the middle. And one side goes running towards the beginning from the middle. And now you have sort of a middled out solution where, hey, maybe it works, maybe it doesn't. But both, like, which side did we start from?

[00:12:27] And so I think the biggest thing that I tell people in terms of what are they doing with AI right now? And how are they thinking about, like, what should I be doing if I'm laying out, like, I'll call it like a course plan. First thing I think people should do is focus on the AI ethics. Like, how are we thinking about this problem? How are we going to use it to address this problem?

[00:12:52] Are we going to have any adverse impacts if we implement something like this? What are the questions that we're going to ask of our tools? Do we need to build or buy? Start with that AI ethics part before you even start to build anything. And then ask yourself, well, okay, is it that our job description? So if you do that, it's like, are our job descriptions really optimized for what we're trying to say?

[00:13:23] Are our candidate touch points as optimized for what we're trying to say? Is it better for us to still continue to train RCs? Or is it better for us to go a route where we have fewer RCs, more agents on this side? And do we effectively trust those agents to be able to hit on things?

[00:13:43] That ethics side, that thinking side, that philosophical side, I think opens the door very differently to conversations that are equally important conversations that will help guide you with what am I trying to build and the design of what you're trying to build. Sure. Yeah, I think there's such an urgency and pressure.

[00:14:10] I would say there's like this productivity promise pressure that talent leaders are feeling because there's an expectation that we're doing something with AI and we're going to get better results somehow. Different companies measure different ways, but a lot of them look at it in terms of what kind of costs are you going to drive out of the business? How is implementing these tools going to help drive costs down? Maybe it's increasing the capacity of the recruiters that could take on more recs.

[00:14:38] Maybe it's headcount as you knew before, but whatever it is, the pressure to deliver some kind of results is there. So there's kind of a rush to just start doing something so they can at least get down that path and hopefully start seeing some of those results. Yeah, yeah.

[00:14:58] I mean, it's I think it's a classic problem that recruiting has always had on because to a lot of foundational leaders and I think there was a post this week where somebody somebody had talked about maybe it was a CEO of a company. So like I think maybe it was like TA or HR is irrelevant. We were going to optimize for it or something like that.

[00:15:22] It wasn't Dario at Anthropic and it wasn't the Perplexity CEO, but it was like somebody else in it like wildfire a post on LinkedIn. And I think that it all boils down to recruiting and TA are in a very interesting spot. We don't exactly fit with the HR side, which is arguably a little bit more compliance and people retention and things like that. We don't really fit in there.

[00:15:52] We also kind of sort of never really fit in with sales. So where does that sort of leave the design? And I think that a lot of people have viewed for a very long time that recruiting is order taking. There is a role. Go and fill it for me. That's the only time I want to have a conversation with you.

[00:16:12] But the true leaders of companies who have built with kind of an HR TA in mind and have held that through through their entire companies, I think are better positioned right now to be leveraging AI tools because they don't see recruiting as this function that is just there to do one thing.

[00:16:34] They see it as a function of, okay, from a business health perspective, how do we recruit? How do we retain our people? Like whose responsibility is it to work that relationship very deeply? And I think that that is, there's something to be said about that in terms of the ROI that then people expect.

[00:17:03] Because I think that a lot of people in TA leadership sort of are still kind of holding on to like, I have to show immediate impact within 20 days or 15 days or anything like that. So I'm just going to go and I'm going to pick a bunch of AI things and roll those things out because I'll look and say, hey, look, we did this. But ultimately, it's the people who are able to sort of say, hey, from a business perspective, we need to change these things.

[00:17:34] And like, we're going to be very thoughtful about it. Yeah, your TA function is so critical to your brand, your employer brand, and how people see you. Yeah, you typically for every job, you're getting dozens. Now it's more like hundreds of candidates because of how easy and automated the process is on both sides.

[00:17:55] But also, of course, the candidate side now, or they can have a customized resume for each job strip to apply for and still hit a button and apply for 100 at a time. So that interaction, it can inform the business so much. And you're right, if companies where your recruiting function is a lot more sophisticated, they are order takers. These are the folks that really understand what the market is for talent, how people respond to your brand.

[00:18:24] And they're the first person, usually, that a candidate, hopefully future employee, is interacting with in your business. So they have so much great knowledge and insight on what the market looks like for talent and how that translates to the folks that end up being your employees, that they really have a lot more of an advisory role.

[00:18:47] And that was a theme I saw in several presentations this week, is making sure that we're elevating into more of that advisory role. And how can you leverage AI to help you in that process? And some of those examples were, like you mentioned, it can help you do better research on not only the job and the company, but the candidate pool in the area that you're recruiting. It's just one small piece.

[00:19:15] And another one is a lot of hyper-focused customization, especially in this very consumeristic world. I mean, everybody expects things to be very customized to them. AI tools can aid in making that a more personal experience, even though it's by using less personal resources like technology.

[00:19:38] Well, it's interesting that you bring that up, because about a year, so there were kind of two posts that I've done. About a year and a half, a year ago, I put out a post that was essentially asking, well, I'll revisit that because my train of thought kind of lost its caboose.

[00:20:00] But I'll come back to the more recent one, which is that I do wonder if, oh, now I remember what it was. Yeah. So we sit on a lot of recruiting in our UTSs, we sit on a lot of data. And one of the things that we always get asked on the recruiting operations side is to pull out trends of people that we've moved forward.

[00:20:26] And the thing that is hard about that is, that's not a very easy thing to currently do, unless you are able to build something to do it. Most ATSs do not have a feature that sort of says, hey, let's take recruiting operations role. Jeremy has moved forward. Tony has moved forward. Thematically, they both have X number of years of experience.

[00:20:55] They both have exposure to SaaS. They both have like, there's nothing out there that tells us that data. You usually just see a funnel and the funnel has sort of pass through rates. And you say we passed, we got a thousand resumes. We moved 200 forward to application review, a secondary application review. We moved 100 forward after that. We short, on the short list from our short list, we decided to interview 50. And you pass it through like that.

[00:21:25] And the thing that I pointed out in my first post about a year and a half ago is something that people are talking about now, which is, okay, from a human component, if I told you, you have two options of interviewing with this company. I can send you an AI interviewer. And you could interview five minutes after you apply. Or you could wait up to two weeks and maybe we'll interview you for the role.

[00:21:54] Naturally, the human response is going to be, I want to get, I'm not applying to wait. I'm applying to get a job. So I am going to take the AI interviewer because it is more beneficial for me to get this thing done now. Okay. That solves for that use case. The second one in terms of customization was about two weeks ago.

[00:22:19] I was thinking about it over a weekend, which is as recruiters become more enabled with AI and have better access to knowledge and have better access to the training and stuff like that. And you are doing interview transcriptions and you've got feedback and how they write feedback. There was all this psychometric data that used to be acquired like 15 years, like 10 to 15 years ago.

[00:22:48] That nobody really has done anything substantial with right now that I think, or at least how I've seen. And you could very easily take that data and say, you know what, do we really need to assign multiple, like one recruiter to funnel one role? Or could we have all of the recruiters be able to recruit on all of our roles? And if that is the case, because we still want everybody to have a human interviewer or something like that,

[00:23:15] it doesn't look like Jeremy's personality from like the words that he's used, the everything that he's done, the responses that maybe he's given. He might not be a good match to talk with Bob, the recruiter. But based on Tony, the way Tony writes feedback, the way Tony writes process, like does this stuff, Jeremy is actually going to be a better fit for talking to Tony.

[00:23:41] Jeremy, instead of having one interviewer like that, we're going to match Jeremy with the right interviewer that's going to bring out what we need in this interview. And I do think that that is a possibility for the future. I don't know if we are there yet, but it is certainly something where I could look at it and be like, that's an effective use of AI because you're bringing out talent in a different way that's not been done before

[00:24:07] and has maybe even sort of been kind of woo-woo and not been an effective use. Yeah, I do think, I think that's a really good point. Sentiment analysis in general, I think is growing as technology can kind of absorb this massive amount of data. Recording so much more. I mean, there's so many meetings that I'm having now that always have bots in there.

[00:24:32] So how do you make sure, not just in taking notes and summarizing later, but how can you apply that technology to map those kind of trends, understand and do an additional analysis on how effective different interviewers are, the way people react to different kinds of questions. What kind of energy in general do they bring to your company when they're interviewing? There's so much that's interesting and companies that are putting together AI interviewing resources and tools

[00:25:02] also have the ability to do that in aggregate and give companies a benchmark, an industry benchmark as well. And I'm glad you brought that up because my next question was going to be, as we get kind of near the end here, what do you see in the whole thing now in the world of talent and technology? And what do you anticipate? Where do you see opportunity to have an even bigger impact by leveraging more tech in TA? Yeah.

[00:25:32] Well, and I do want to touch on one thing from your response that I think is kind of interesting. You know, it's like right now when we talk about scheduling, we talk about interviewing with people, there's this sort of reliability on people to say, hey, you know, I'm open to doing two or three interviews a day and I do them, you know, book me in the morning and stuff like that. And to your point about having all these conversations right now,

[00:25:57] I think we can actually get to a point where we're using it to be actually effective and say to like somebody like, hey, I know you said that you can do four interviews in a day. You know, Tony, that's great that you're that gung-ho about it. We mapped the 20 interviews that you did this week. We looked at what it was for you in the morning and we looked at what it was for you in the afternoon.

[00:26:24] And we looked at it from one to like, we pick a different set of time and we say, you know what? The questions that you asked, the engagement that you had in the morning was very different than your engagement in the afternoon. Can you tell me a little bit more about that? Is that a trend for you or something like that? Or maybe we take the bulk of the data and we go and we look at it and we do that sentiment analysis. And then we come back to you and say, hey, Tony, look, love you as an interviewer.

[00:26:53] Love the energy. Can we be frank with you? After about two o'clock, nah, you're not a good interviewer. You're just not yourself. And your assessment of these candidates, like you're not paying the price. The company is paying the price and the candidates are paying the price. Like how do we fix that? And maybe that is the answer to your question about where do we get the biggest impacts?

[00:27:23] I'm also pretty bullish on we're going to have agent to agent interviewing. I think that that's coming a lot faster. I think you see people talking about building their digital twins. I think the evolution of building the digital twin is I have my agent. Company has their agent and hiring manager has their agent. And it's like you cover three interviews in a nanosecond because my agent's trained on everything for me.

[00:27:53] Recruiter agent, hiring manager agent, technical agent, if you need one. They know what's going on. And then you have sort of a, all right, we've just dropped time to fill down to like really single digit days because all these interviews have happened really, really quickly and we have high signal on all this. And there we go. And that's huge impact too.

[00:28:20] I don't think you take humans totally out of the loop because you need them there. But I can see a world where people are just more comfortable with, hey, agent doesn't get tired. Agent is going to be consistent. Obviously, I think AI interviewers now are essentially doing what we've always known, which is structured interviewing is a better thing to do.

[00:28:47] And we can't get that out of people, but they are going to do that 100% of the time because we've told them to. So I think that there's a lot of impacts to be felt. I think that everything is still evolving very quickly. I think that we're still in a spot where we're trying to figure it out and we will. Yeah. You know, that or the, what is it? Mythos is going to replace us all and find all the zero day things.

[00:29:13] And all of the, like somebody's going to figure out how to embed a something in the back end of their resume. And that essentially will give them feedback or expose the, you know, ATS side feedback so that they can see where they're at. And, you know, know who looked at their resume or if their resume was even looked at. So I think that, you know, there's, it's, it's yet to be seen. For sure. But I agree.

[00:29:44] I already see how easy it will be for at least the first couple rounds of, you know, trying to find that match between a business and a candidate being done by robots. You know, while everybody else is sleeping. And to me, that's great.

[00:30:03] I think it's for a lot of folks that can seem a little bit intimidating, but anybody that's been through the interview process, probably everybody that's listening has been through an interview process. They know it's a lot of work. It's a lot of time. And especially if you already have a job, that makes it really difficult to just keep up with the process. So if some of those things are happening for us, representing us well, representing the company well, and they can go on without our involvement. Great.

[00:30:32] To your point, there's still going to be a human at the end of the day that's going to talk to you and, and, and help you make that connection to the company. So, well, and I want to, I want to do, I do want to ask you this question before we sign off, which is like, all right, so you're given the magic wand. You can see 10 years into the, like, you're given a magic ball. You're given a magic wand. You will, you're able to see 10 years into the future around what it, what it all looks like.

[00:30:56] You can wave that magic wand at one thing right now that you're going to fix with, with this technology. What are you doing? Yeah. And why? I think it really is along the lines of what we're talking about. I could, I could actually envision a day where, you know, my daughter's getting a job. She's, she's about to turn 12. She's getting a job in the future and never goes through the interview process.

[00:31:25] All of this stuff is taking care of the, doing all the matching. Her skills are probably verified, which I think is also a rapidly growing, a rapidly growing and evolving trend. And, um, but I think it's, the matching is going to be almost automatic. So in even the negotiating, you know, the, everything's going to be, is becoming increasingly transparent and increasingly frictionless.

[00:31:50] So I think it's more likely that she'll be like, I can't believe you used to spend days, you know, going through all, and hours and hours, going through all these interviews, getting dressed up with a tie and, and going to these businesses instead of just mashing and starting to work right away. I just think it's going to be a lot more fluid and immediate. And that's, I think that's what a lot of people want is speed to talent. So the faster we can get there, the better. And it's going to have a lot less costable. Yeah. Yeah.

[00:32:19] No, I mean, I think, you know, the future is, future is our oyster. It'll be interesting to see what happens. Well, listen, folks that are listening to this, how do they get in touch with you? What's the best way to get in touch with you to learn more about RecOps, about your community and how you can? Yeah. So best way to reach me is reach out to me on LinkedIn, Jeremy A. Lyons. If you want to shoot me an email, it's Jeremy at RecOps Collective, spelled exactly how it sounds, .com.

[00:32:49] And, you know, I love getting messages. So just mention that you heard me on this podcast and I will, you know, that way I'll know how we got connected. That's great. Well, thanks again, Jeremy. Thanks for joining Human Cloud. And for all of you listeners, please go check out Jeremy's profile. Check out his posts. Learn more about the RecOps Collective. Thanks for joining us. Thank you.