Ep 834: Six Predictions For AI In HR

Speaker A: There's no shortage of predictions about what AI will do to hiring and hr. Most of them come from people watching from the outside. It's a lot rarer to hear from someone actually making the changes inside a company that's already living this. So what does that view from the inside tell us about the year ahead? Keep listening to find out. Support for this podcast comes from Mackie. Mackie began by replacing the resume screen with a fair, structured voice interview that assesses real skills before anyone formally applies. Now that same intelligence is extending across the whole funnel, from first conversation to final interview, giving hiring teams the insight they need while every decision stays in human hands. They recently announced Tomo, an AI interview assistant for hiring managers. The next step towards one connected system that screens interviews and gives every candidate a consistent, fair experience at scale. See how the end to end picture comes together by going to mackypeople.com that's mackiepeople.com and Mackie is spelt M M A K I I.

Speaker B: There's been more of scientific discovery, more of technical advancement and material progress in your lifetime and mine than in all the ages of history.

Speaker A: Hi there. Welcome to episode 834 of Recruiting Future with me, Matt Alderman. There's no shortage of predictions about AI in HR and hiring at the moment. Most of them come from analysts, vendors, and commentators looking in from the outside. Far more useful and much harder to come by is the perspective of someone actually doing the work, running the transformation, and making these changes inside their own organization. That is exactly why I wanted to talk to this week's guest. My guest is Vika Kosole, Head of AI Transformation and People Strategy at Cloudflare, where he also serves as Chief of Staff to the Chief People Officer. He sent me a set of predictions for AI in HR in 2027, and they make a great framework for understanding what is really changing and what it means for the people doing this work. In our conversation, Vic talks me through his predictions and what each one of them looks like in practice. Hi, Vic, and welcome back to the podcast.

Speaker C: Hey there, Matt. How are you?

Speaker A: I'm very well, thank you. And it is great to have you back on the show. Now, you were, uh, last on Recruiting Future back in 2022, and we were talking about People Analytics. So, uh, tell everyone who you are and what you do now.

Speaker C: Sure, sure. Well, first of all, Matt M. Glad to be back. Thanks for having me. So, by way of introduction, Vic, Vicky Kasale. I am, um, the head of AI Transformation and People Strategy for Cloudflare, which is a AI infrastructure and cybersecurity company. I also have a dual hat and serve as the chief of staff to our chief People officer as well. That's the people strategy half of that role there Matt. So um, still same data driven, evidence based driven work just now embedded in that is AI transformation and how do we evolve our own future function and then more so how do we help the business transform?

Speaker A: So there's not that many people doing that kind of role at the moment. So I'm sort of fascinated to uh, talk to you, I mean tell us a little bit about what does that involve on a day to day basis and also you know, what are you learning about where AI is working in HR and where it isn't.

Speaker C: Yeah. So I think there's really kind of four key components of what that looks like on a day to day. One is AI strategy, what do we do, what don't like we do why and what that's really like the first big piece. Second is AI governance. Right. And I think of governance of how do we enable people to do the good thing, the right thing more often and then make sure the bad things don't happen ever. Right. AI governance and then three is a really really cool space of product development and that's an area that I've been trying to champion that more and more HR leaders, people, analytic leaders, recruiting leaders like they need to become or needs to at least adopt more of a product management mindset. So that's that third like we made it really the fourth one is uh, all other duties as a sign that comes with being the chief of staff person. So a day to day that can look like I am leading a transformation workshop to help us rethink how we approach a certain workflow. And then in the next meeting I'm getting updates of um, a new product that we've wrote up because there's some feedback. And then the next meeting meeting I'm doing some training. In the next meeting I'm uh talking about how do we rethink about workforce planning, headcount planning in terms of is it an agent, is it a person? Is automation. Right. So that's an example of what VIX calendar looks like. Not saying that's exactly what my calendar looks like today but just saying I got that from somewhere.

Speaker A: So you sent me a uh, set of predictions for AI in HR in 2027 which I thought would be a great way to frame this convers and also reminded me that we're recording this in October so we're heading towards that it's that time of year where these things start happening. So we haven't got time to talk through all of them. They're all really, really interesting and I know they reflect things that are, that are happening in your uh, sort of in your organization at the moment. But let's just go through six of them. So the first one kind of comes back to something that you've just said, basically. So it's the prediction that more HR teams will build and maintain their own AI agents and applications. Now this is a big shift for lots of organizations because HR has traditionally not been particularly tech savvy and has bought its technology from, you know, external vendors. So. So what's changing? What's happening here?

Speaker C: Yeah, so, um, a bit on that. I think most organizations and most folks don't think of HR as being a high tech organization. But one of the benefits of being chief of staff is I see the budget and we spend a lot on technology, so we' really high tech. It's just we've outsourced like a lot of it. And unsurprisingly, Matt, what's changed is that AI changes the, the equation of do we build reverses by to. There are certain things that sometimes it may make sense to build where the default was. Just we need to look for a software application that does this very specific thing, right? And the incentives and economics of building a SaaS empire or you know, a business aren't the same as maybe a very niche problem that your specific business has. So more and more AI, because it can translate intent to language and code is a language, we can now begin to build our own internal applications. Now the challenge is that what happens when everyone starts to do that, okay, uh, okay, all of a sudden we, instead of outsourcing that we are now bringing on that engineering cost, that maintenance cost, all the things that we used to outsource, we are now in sourcing and there needs to be some strategic product management around that. And more and more people will start to think about that equation a little bit differently.

Speaker A: I think that's really interesting because the kind of debate around this m at the moment, it's very, very kind of unnuanced. It's just like, well, people are going and building their own ATS systems and that's ridiculous. But actually, you know, what you're talking about here is some very, you know, niche application specific problems. And I think anyone who's experimenting with some of the frontier models on AI will know, you know, how quickly and easily these kind of problems can be solved by building, just building something in, you know, in a couple of hours or something like that. So it's a really interesting area. And to dig into what you said there about someone to manage all of that, your second prediction is that HR functions will appoint internal AI product leaders. Tell us a little bit more about that and in particular the skills that that person is going to need to do the job.

Speaker C: Yeah, yeah. So I think there will be a new role, a new executive role coming within a lot of functions. But specifically, let's talk about hr. That there will be a. Let's call it a VP of internal AI and product development or internal product development and AI. And why will that role need to be there? Well, going back to my previous prediction is that there will be a spurry of people who are building all the things. But what happens to it maybe after. So here's an example to cut myself off, but here is a example that I think is really, really fascinating about. We are now working and managing agents of people who are no longer at our company.

Speaker A: Wow.

Speaker C: Yeah,

Speaker A: I need to absorb that. So people have built agents that.

Speaker C: People who have built agents. Right. We have

Speaker A: no longer

Speaker C: cycles enough. Right.

Speaker A: And their agent is.

Speaker C: And their agent is there or what they built is there. The applications are there, but they are not. So uh, we're working with the agents that they build. How do you manage all that? Who owns that? What's the life cycle? Right. That is not a core function or core thing that we use to be that we used to have to think about. That is a new problem.

Speaker A: Yeah, that is a new problem. New problem.

Speaker C: So that's why this new problem will require a new leader to lead it. And what that role will really need to understand is evaluation. Is this working? Is it not? Right. A good sense of product judgment. Right. Just understanding, not necessarily how to be a software engineer, but the architecture that goes into making really good product usability. User experience is a big one. But I think most importantly understanding how to pull from people the processes and logic that's in their head to codify that into agents, applications or automations that now the business can use and then create systems, programs, life cycles to manage it whenever whoever built it may have left.

Speaker A: Yeah, I think that's really interesting and perhaps something that m people haven't really considered. I mean it's such an interesting. It really is such an interesting area. Your third prediction I think speaks to something that we've seen really play out over the last couple of years. So your third prediction is that businesses are going to judge AI on um, much more than just time saved or cost avoided. And I think this is interesting because to date they've been the things that people look at and you know, people haven't got the results they needed. They've perhaps restructured their business in the wrong kind of way. What, what else should people be looking at, uh, in terms of business value from, from AI? And how does that build into a business case to, to, to use our implementer?

Speaker C: This is probably one of my favorite topics to talk about the ROI of AI. One, lots of levers, alphabets, uh, but two, it's because, because AI was a new financial item in everyone's budget. It's a new line item, token cost. We never had token costs before then. We didn't understand what the value of the token is. So a lot of times when we don't understand the value of something, we default. Call it our core ways of understanding value. Does this save me time? Do I make more money? Here is an example. For a lot of HR functions, the most expensive technology they buy a lot of times is their HRAS system. No one asks what is the ROI of our workday platform. Right. Because we understand that the value that we get from that isn't just financial or time saved, it's compliance, it's experience, it's quality. Right. There's lots of other things that we know. Having a system like that but brings us a value. AI, I think it's going and it's still in it's kind of hyped cycle. But my prediction for 2027 is that AI will start to become more embedded into work and we will not start to think about what's the ROI of uh, like a token, but what's the value that we're getting. So there are six or seven different dimensions of value that I always talk about. One is efficiency, which is just time. Two is scale. Right. Does what we're buying allow us to scale? Right. Great. Three is quality. Does it improve the quality of what we deliver? For that is experience. You know, does this improve the overall experience to the employee? Then we have risk mitigation. Does it reduce risk? Does it help to give us a new capability that we did not have like before? And then finally when you look at all of that, you can translate and connect any one of those to the financial impact it has. So yes, we will start to look at AI value very differently now.

Speaker A: That's a really interesting one. It's something that people really need to think about because AI also, it almost becomes like Infrastructure. It almost becomes like an Internet connection, doesn't it?

Speaker C: Correct. No one's, what's the ROI of my Internet connection?

Speaker A: Very true. It's almost a

Speaker C: comical question to ask, right? But it's about, well, what do you do with it? Right. Which is where a lot of our, uh, questions will start to shift. Because AI is here. It's not going anywhere. It's about, what do you do with it? And how do you convert what you do to value.

Speaker A: Now, your fourth prediction is an interesting one, and it's something that we're certainly sort of seeing play out at the moment. So it was that employees who are using AI themselves and getting gains in how they work. If them working faster with AI just means they get more work, then they're going to kind of hide that and not tell their organization that they're using AI. And I'm kind of seeing this. I know a number of people who've automated large parts of their job and haven't told their employer through fear of that and a number of other things. How's that playing out? And what should HR be thinking around this?

Speaker C: Yeah, I think this is probably my most controversial one. If AI helps people do work faster, easier, better, sometimes the reward for work well done is more work. And there are two sides of that argument. Well, now you have more capacity, you should be doing more stuff, right? Or it's the side of, hey, this is the work life balance that we've all been dreaming of. So cut us like a break here. And I think if you think about what typically happens and just play it out from a behavioral standpoint, this is inevitable if certain things tend to happen. So here is what I think when I talk to folks and colleagues were already saying, if you can do more work with AI, then you will get more work to do. Now, more work to do does not, uh, necessarily mean that the cognitive load of that work, especially for knowledge workers, decreases. Right. Personally, if I'm working with AI, I can work on eight things at once. The question is, Matt, am I doing any of those eight things exceptionally well? Am I focusing on what I'm actually doing? Or am I missing a few things by just essentially scroll syndrome, where the next ping, the next ping, the next ping. I'm jumping from screen to screen to screen. And if there isn't a methodological way to understand not just output load, but attention load, right? Then that is what is going to be an issue. And that employees will start to say, hey, you know what? I can do my work, but I don't have to tell you how I'm actually doing it.

Speaker A: Support for, uh, this podcast comes from Pebble. I know that many of you listening are hiring internationally and you'll be all too familiar with the wall of paperwork, rules and approvals. And you know just how much this red tape can slow things down. That's what pebble is designed to fix. Pebble makes global hiring simple through embedded compliance and AI driven workflows. The pebble platform takes the delays and guesswork out of going global. So you can move fast without adding risk. The bottom line is that anywhere is possible with Pebble. With global hiring simplified, founders and HR leaders can spend their time on focusing on what and where comes next for their business. Pebble has a special offer, uh, for recruiting future listeners. Pebble is normally $399 a month per employee, although a no brainer for what you get. But right now there is a limited time offer on their site that makes it even easier to get started. Go to high Pebble AI before it's gone. That's HIPEBL AI. Terms and conditions apply. Yeah, it's such a big consideration that people really need to, really need to think about. So moving on to your fifth prediction, which is something we've touched on a little bit already, it's that work design will come before headcount planning. Explain what you mean by that and what it looks like in practice.

Speaker C: Yeah, so headcount planning, uh, we typically traditionally the way that's done is we say, hey, who needs more headcount for what? Right? Or uh, for what do you need more headcount for? Or we said, hey, we need to do something and we need somebody to do it, therefore we need headcount, which is great. That is, I think, the traditional way of doing workforce planning, headcount planning. But now because we have different tools that enable us to get work done, it's less about the person and more about what is the job that needs to be done. Right. So can we describe what the work is? Oh, okay. Well, the work is somebody needs to be enabled to respond to employees quickly, fastly. And we plan to triple our size next year. So we obviously need more people. Oh, okay. Well that is a job to be done. We can bring in more people. Or again, and this is like the part that I think folks get wrong, we can bring in a certain kind of person to help to build the AI infrastructure, automation for that work. So it's not that headcount planning means that you hire less people because people think, because you can just, you can Hire an agent or you can substitute headcount for an agent. We're finding that it's actually more nuanced than that. Even if an agent can do the work, somebody needs to be responsible, or somebody needs to build the agent, maintain the agent that does that work. That's still headcount. But maybe on the other side you get a bit more leverage on what that person can do. So maybe they build an agent for this team and they build an agent for another team and another team. But again, there's a capacity to maybe how many agents that you can really, really manage because they're products, if you think of them as products. Right. So that is where I'm thinking that the world will start to shift to say, well, let's not start with the headcount. Let's start with the work that needs to be done and then understand the various ways that we can meet that demand for work that needs to be there. Because now we have a different pool of supply, if you will.

Speaker A: And sixth prediction is sort, um, of particularly relevant for talent acquisition recruiting. Recruiters will use more work samples and live exercises to assess judgment. It's interesting because judgment has been something that we've been talking about a lot on the podcast in recent episodes. Give us your perspective on it. Why has it become more important? What should tele acquisition be doing about it right now? Really?

Speaker C: Yeah. So right now I think that most people know that we are in this AI echo chamber when it comes to recruiting, where companies are using AI to either, you know, score or rank candidates and employees are therefore using AI to ensure their applications are ranked highly via the algorithm. So everyone is using AI and it's like employees or candidates, AI are talking to employers AI. So how do you know that somebody either has the skills, knowledge and abilities to do the actual work? Well, that is the space and the problem, uh, set that I think recruiting and TA functions need to solve. One of the ways to do that is kind of just old school. Let's talk about, or let's have you work through a problem life. Right. That is one of the best ways to just kind of see how somebody thinks. There's also a element of verified work history, because I know that's a big thing. How do you know that somebody's actually work at the place like they've worked, but when they're across from you, across like the screen. Right. How do you, uh, assess them? And I think the best way to assess is to see what they can do live, give them a project in the moment. The project is small, but at least finding ways to assess capabilities live will be a natural default. Now, is it the best option? That's what I don't know. But it seems, you know, I like to say, well, where do people typically tend to go down? And I think this is a way that people typically will tend to go down. But let's see, by this time next year, if I'm right now.

Speaker A: Yeah, I think. I think there is a big rethink coming in terms of what it is that companies assess people for and how they assess them. And it will be interesting to see how that plays out, how quickly it moves and where we get to. So, uh, yeah, I'm with you. I'm interested to see what that looks like in a year's time to sort of sum up a little bit, because obviously a lot of what we've been talking about is kind of pretty technical stuff. There are people listening who aren't working in technology companies and sort of building and running software. Not really traditionally been part of HR's job, just obviously spending lots of money buying it. Where do people need to start with this? What should they be thinking about? What's the first move that people should make?

Speaker C: Yeah, great question. In fact, this is a question I get a lot, actually. And it is essentially, how do I become an internal product leader? Like, what's that gap between where I am now to where the business will need me m to be? And I think it really comes down to three core skills that some folks may have and then some folks. It really isn't technology driven. It's just, how do you help people think? So the first thing is helping people to distill process right into documentation. That's not technical at all. Yeah. And that feels like, huh, huh. Is that really a core component? And the reality is, yes, I can tell you so many times I've been trying to work with teams or partners to leverage AI to change how they work. And, you know, we have a conversation and we get to the point where we realize we're not really clear on how the work works here. We need to really outline what this process looks like. The happy path, the exceptions. Right. That is a facilitation skill. That is a process engineering skill. So for all folks who thinks that I need to become a software engineer. No, this is the number one skill that you need. Okay. Skill number two is just a general understanding of what AI is good for and what it's not good for. Right. People talk about AI, uh, as a broad kind of monolith now, but I like to break it out into maybe kind of three to four pieces. One is a assistant, so something that chats back at you, you know. Second is a automation, which is how do we move data from point A to point B. Three are the cool applications, that's the third A that folks like to see. And then four is an actual AI agent. Right. So understanding when one of those four is a requirement is I think the most that uh, folks need to understand. And you're two thirds of the way there. And that has nothing to do with a technical capability. Okay. The third piece is really what I call product judgment. And that does require you not to understand how to code, but just understand all of the pieces that goes into a good product user experience. As I was saying earlier, where does the data live? Is it living in a system of record or do you have to create a new database? So this is more architecture of how do all the pieces fit together, then how do I write the code to actually do it? Because if you are a leader in this function, writing the code is no longer the hard part. AI makes that sectionally easier. It's the product judgment. Knowing what to build and why does it actually solve a problem. How do all the pieces fit together? That is a core piece there that I think if folks are looking at how to transition, that's a one, two, three step process of things that you may need to know.

Speaker A: That's very useful. I know that lots of people listening will find that really valuable as a, ah, final question. We're obviously talking about predictions for uh, 2027 and also a lot of things that are happening right now. Let's just look a little bit further out. So what do you think the future looks like beyond that? Where? HR talent, acquisition talent. Where are things going to be in two or three years time, do you think?

Speaker C: Yeah, yeah. Ah, it's hard to predict more accurately beyond next month these days. Yeah,

Speaker A: that's very true. Next week sometimes.

Speaker C: Exactly. But here is where I believe things are going. I think AI will have its hype cycle. And I say hype cycle not in the terms of it's all hype, but hype cycle. That it's the thing that people are fixated on. Right. Right now people are very much so fixated on how do I use AI to do X. I think the hype cycle will cool down and people would just say, well, how do I just do a better job at this thing? AI or not, AI is one tool that I can use to solve my problem. But how Do I better solve the problem and bring all the tools that are available to me to solve that problem to bear? And, uh, I think AI will fit into the context of that. Now, I do think where I believe the world is going is that work will be very, very different, especially for knowledge workers. But I think work is all around where you will have your own AI work agent that is assigned you, just like your laptop assigned to you. That agent will be able to access certain things, do things on your behalf, get approval. In the same way that you have a, uh, single sign on identity for your work, your agent will be authorized to log certain systems. Right. So that is where I know the work is going. I see things like, you know, Meta released Muse as a personal AI agent. There's things like Rockbot as a personal AI agent, dots from OpenAI. And what tends to happen is that a lot of these personal tools, they're looking for a enterprise use case because that's really where they make a lot of their revenue from. So the writing's on the wall here, Matt. We were, we are all going to have our own personal agents, but it will become so normalized by that time because of just how prevalent it will be that it would be a natural transition and not so much of a culture shock in the same way large language models were. So there you go.

Speaker A: There's some great thoughts there. Vic, thank you very much for talking to me.

Speaker C: Yeah, I always love being on. Let's not make it, uh, four years next time.

Speaker A: Absolutely. Yeah. At least next year. Yes. To see how much of this happened. Yeah,

Speaker C: Indeed.

Speaker A: Great to talk to you. Thank you. My thanks to Vic. You can follow this podcast on Apple Podcasts, on Spotify or wherever you listen to your podcasts. You can search all the past episodes@, uh, recruitingfuture.com on that site. You can also subscribe to our weekly newsletter, Recruiting Future Feast, and get the inside track on everything that's coming up on the show. Thanks very much for listening. I'll be back next time and I hope you'll join me. M. This is my show, Sam.