Latest / Range Podcast / How Compensation Teams Should Use AI
Transcript
- Giac Soliman: Welcome to Range. I'm Jack Solomon and this is the podcast for competent reward people building with AI without losing their grip on the pay decision. Every episode, one practitioner, one honest conversation about what AI can actually do in our world and where you still need a human in the room. Be curious, stay a little skeptical, and here we go. Giac: Good morning everyone and welcome to a new episode of the Range Podcast. Today I'm here with Arif Ender, who's a highly experienced SolarWars professional, has been around for quite some time. his career spans almost two decades in multinational corporations, such as Palo Alto Networks, Meta, Mars, and Esle. He's currently the director of compensation for EMIA. And Latam at Palo Alto Networks, where he leads compensation strategy and program implementation across brace for it, over 65 countries. Arif, thank you so much for joining me on the podcast today. Tell us a little bit more about when you started using AI in your work. Like you've been around for two decades, if not more, right? you've been doing the work in the more traditional way and over the past four or five years, you know, this buzzword came around, AI, and you probably looked at it, it looked at you, it was a match made in heaven. But tell us a little bit more, how did that happen? what was that moment where you decided I'm gonna start exploring this? Arif Ender: first of all, thanks for the opportunity, Jack. It's always a pleasure, you know, talking to you. so where did things start? I had the same question last week when I was attending a seminar, a rewards seminar here in Amsterdam. Technically, if I really look back it was when I first joined Facebook. No, I mean now it's meta, it was Facebook when I joined. And I got one of the first tasks. it was about like an sentiment analysis that was done manually. you know, about these pulse or engagement surveys as they call it. And we wanted to get a kind of a understanding about like how employees feel about compensation and benefits at at Facebook. So and I Giac: Mm-hmm. Arif Ender: said like there is there might be a better way to do that. And when I was talking to the to the other it was just because they were also Turkish. I was also trying to make some you know connections. I was fresh in the company and some of them were really just working on machine learning Giac: Okay. Arif Ender: to feed feed our platform security, you know, they were just basically using some posts and stuff to kind of feed that in and train. And I said like, ooh, this might be cool actually. I can use the same technology, maybe just do a language processing. and I got access, I asked for it, I got access to that platform. Do the machine learning and I basically uploaded I don't know how many, still it was like hundreds of comments, and Giac: Right. Arif Ender: it was like boom. It just kind of proposed like you know how many is positive, how many is negative, but it was in literally minutes and I would say if I will do manually, I would say like an eighty percent accuracy which I buy it, you know, for Giac: Mm-hmm. Arif Ender: the for the time gain. And that was the first time I was exposed. Technically what we call is generative AI, you know. Giac: Gotcha. You seem to have some familiarity with the terms and the technology. Out of interest, have you always worked in compensation? And was that a like a deliberate choice you made when you started your career Arif Ender: how I started is also interesting. I studied engineering. By education With a master's in industrial engineering and technology management. so there comes the the geek part, I think. Giac: Yeah. Nice. Arif Ender: and I also did software development while I was studying like a part time. while I was at university. So I started as an engineer in Mercedes Benz. Two years later I realized I think I need more exposure to something more people centric. Although I enjoyed studying engineering, I think that analytical thinking, critical thinking, whatever those skill sets now, the STEM, what we call next STEM education, I truly believe in that. I enjoyed it. But in the real life I wanted something that relates or connects more to people. And basically I went back to HR. You know, I said, I want to per pursue something else. What can we do? They told me sales, marketing, no, we also have this opening in compensation and benefits. Said, okay, that seems interesting. What do they do? I had no idea, by the way. Yeah, basically it's all about managing people how are we gonna manage pay, what we provide as benefits. Said, hmm, interesting. This is pretty close. Yeah. Yeah, I had no access to that. That's the thing. It was Giac: You're you're there thinking probably, it's it's probably a high paying job, but since it manages pay, you know, it's like I'm gonna go in there. Yeah. Arif Ender: just an assumption. But when I look at the team, interestingly, my manager back then and all my colleagues, they were all mechanical or industrial engineers. I said, Okay, yeah, this might be a good fit, you know. Within HR, it's the more engineering type part of the HR. So let's give it a try. I mean, worst case scenario, you know, in Turkey there's this compulsive military service. Worst case scenario I go to military. You know, I had to do it anyway. So Giac: well glad you're you're not in the military right now with everything that's happening in the world. like Arif Ender: No, yeah. Giac: that would be a challenging situation. But it's interesting, right, 'cause you look at other leaders working in compensation, total rewards, we all have different backgrounds. There's some people have an economist, like are economists by background, some people are lawyers by background, some people are, you know, they they've studied philosophy or physiotherapy, so diverse. You know, I know that in your blog you mentioned an identity problem that could, you know, impact some of the reward teams out there. And I wanted to learn more about it and see what your thinking is in terms of how AI is reshaping the perception and identity of the compensation team. Arif Ender: I'm sure a lot of the rewards people will correlate what I'm saying. Basically, I think now the terminology or basically how we identify ourselves should change. if you really look at it, the things that we build or we are building in our current profession or what we are managing, they are We always look at it like it's a program, you know, compensation programs managers. Or it is like a a project management, a big project coming up, we're gonna do, let's say, an incentive design. But technically, I think we should start looking at it from a product management perspective. First of all, technology is shaping all the professions, not only ours. And secondly, we are going towards that mindset. You know, like you have developers, you have product management. If you if you really look at the software world if you look at a product organization in any tech company, or even in non-tech company, basically the R and D part, you will see there is developer operation like DevOps, then you have the builders, software engineers, and there is these product managers who who manage the whole roadmap. Giac: Mm-hmm. Right. Arif Ender: I think we were already doing it, we were just not calling that one out loud as a product manager. And then I think it's time that we might need a mind shift towards that. So technically rewards Giac: Yeah. Arif Ender: people, I see them in the future as product managers. Giac: Yeah, we've been talking about bringing product thinking into compensation and more broadly HR for quite some time. But there are just a few companies that were able to successfully do that. But I think now with AI we've really pulled a curtain on that debate. And speaking about debates, I'd love to learn where you stand. Should every compensation professional become a product thinker, AI builder? So by that we say we mean using AI automations and no code tools like Codex, Cloud Code, Lovable to create automations, to create workflows and even perhaps internal comp apps for specific use cases. Where you stand on that? mm-hmm Arif Ender: there are a lot of prereqs, let's say, some requirements Giac: Mm-hmm. Arif Ender: to be able to do that. First of all, if we are giving, let's say, an analogy from manufacturing, you know, the Kanban methodology, first in, first out, right? And if we kind of also translate it to data, crap in, crap out. So If you really do not have a good data infrastructure, if you do not have solid policies that can explain everything, the rationale, the eligibility, you know, basically any guardrails, ring fencing towards any certain program you have. If they are not documented well, if they are not version controlled and stuff, so basically whatever you build on top of it. is gonna bring you way worse outcomes and Giac: Headaches. Right. Arif Ender: at scale because the pace and the speed and the the the scale of AI actually multiplies multiplies h also the worst things. You know, yeah I mean if you give something good it makes it better at scale, right? do the opposite. If if something is wrong, Giac: Right. Arif Ender: that mistake becomes really way more bigger than Giac: Yeah. Arif Ender: if you will if you would have done it manually. Giac: So what you're Arif Ender: So Giac: saying is data, you know, as the foundation of your AI models and getting your house in order before you build is pivotal, is a pivotal moment. Count Arif Ender: Of course, yeah. And and I I can I Giac: Counterpoint. Arif Ender: can give you example on that one. we are in Google Workspace here. So we are using Notebook LM. And the Giac: Mm-hmm. Arif Ender: the one of the ideas we had, like, you know, we are getting a lot of tickets from the recruiting about, you know, what is the allowance in this country, what is I don't know, job to price in that country, is this country eligible for providing RSU? Like all these questions, technically super basic questions with with maybe one policy to be explained or referred to and stuff. So let's do let's do a a library on notebook LM. You know, basically we just upload all the policies, all the tables that is open to TAB and create it as a chatbot. You know, they can go instead of raising a ticket, waiting for a certain SLA, they can literally get it in seconds. They type and not Giac: That's an excellent idea. Arif Ender: yeah. The problem we had one or two Giac: huh. Arif Ender: policies, we we use the outdated versions. And that's where I say, you know, that that mistake can multiply at scale. Giac: Yeah. Arif Ender: and basically we had to clean up so much mess in a couple of offers because the the outdated policy had obviously different numbers, which Giac: Mm-hmm. Arif Ender: were outdated. So, and that went to offers. I was not cool. So that that is the best example. You know, it's super simple. One policy, update it, it can create so much mess that eventually you have to again go back, intervene manually. Giac: Mm-hmm. Arif Ender: And I thought, like, I I wish I I could basically open that ticket manually, respond, and close it. I wouldn't have this. You know, I spent way Giac: Right. Arif Ender: more hours. So that's why I'm always just we are going like one step back and see, okay, is everything in order as a foundation? Then we build Giac: Absolutely. Arif Ender: on top of it. Giac: Counterpoint to that, it is clear that before you start building with AI you want your data house in order, right? Your policy versions named and in in in line for you to make sure to pick the right one. But what if AI can help you with that? Do you envisage a way or a scenario where AI can help you get that data in order. Whether it's your salary data, whether it's your historical performance ratings, where it's your offer premium offer policy or recruitment policy. Do you see a way we could use AI for that to to to make it less cumbersome? Arif Ender: we are in a profession that really requires, I think, an oversight. And the one thing that I I also mentioned in in last week's seminar because it was way more like I will say IMEA specific as well. the EU AI Act is coming up. And one of the biggest Giac: Mm-hmm. Arif Ender: requirements there is the human oversight. So first Giac: Yeah. Arif Ender: of all, we can't let do like let AI do everything. And you know what? okay, it's perfect. You know, technically we should be able to prove that a human being has reviewed the output. So I agree with that because no matter what, I think we are in a profession that there is still somehow a a human influence or interpretation is still required because HR in itself is not like that black and white. Yeah. I mean, at the end of the day coding is, if you even f train a small language model with just one guidance about a certain product or documentation of a language, Giac: Mm-hmm. Arif Ender: coding language, I'm sure they can start right away, you know, coding Giac: Yeah. Arif Ender: in a in a good quality. But HR is not like that. We don't have just one playbook to play. when there is human in it, it is always a grey area. And unfortunately AI is not there yet. I think or I won't risk it, basically. The risk is too much to let it go. Giac: Right. I agree. I I wrote about this as well. To your point, I think that's what the AIU Act wants to tackle. It was it wants to make sure that we do not inadvertently decide to switch off our brains because now we have this new capability that endoubtly can do jobs better than us. Like cleaning up your folders. I I had a go at that too by the way. And you know why I was so amazed? Because Cloud Code Not just cleaned up my folders and reorganized them. It added a README MD file explaining what every folder does for better context retrieval every time I ask Clotcode to, you know, look at my folders, my documents, and do something with it, Arif Ender: you know. A at the end of today, I think AI is gonna become your best ever analyst, comp analyst or rewards analyst that you can have. But still, you still have the anti decision making, the influence that you're gonna have t as a business partner. the brain behind the strategy basically y you'll get the inputs instantly, probably in a way more accurate, in a may more structured way. But who's making the call? I'm glad that you read the blog post. what I'm referring to besides the product management view. Another thing that I'm trying to kind of influence e even in in my team here as well, is basically we are architects, not analysts anymore. Like if you if you are stuck in the analyst definition and if you see that in rewards, I don't think you have too much life span in this profession. But if you see yourself as an architect, that's another thing. Yeah. Giac: I agree. I think what the role will become is like you're not analyzing yourself the data anymore. You are orchestrating different tools or agents that call different tools through APIs, MCPs, you know, etc. to do that work for you, but then you are reviewing everything. You you are essentially the orchestra's director. You're not the musicians anymore. The musicians are your agents, the instruments are the tools they call to do get the job done, and you're directing all of it. So I to your point, I think that's what the analyst role is becoming. It's some something of a architect slash orchestrator. Now there will endoubtly be junior compound list listening to this podcast, Arif. And I wonder what's your advice? For someone that's, you know, maybe being laid off, especially in the tech space. We've seen a lot of laid offs, arguably for due to AI. Who knows if that's the just the narrative of the of the or the reality? But what what's your advice for them in terms of how they could how they can upscale and and and and avoid becoming obsolete in the market? But it's not just for juniors, right? It's also for seniors. Arif Ender: Yeah. I'll say one sentence and I think that that that is a really good headline, if you ask me. I think for the next couple of years or next decade, let's say, let's make it a little bit long term. If you are valued at this point, how you do a great modeling in a certain spreadsheet and this is your value to the company or to an organization, or this is the skill set that you are bragging about. It's gonna it's gonna be really tough. But if you are valued because of your insights, critical thinking, trying to utilize such a technology for your productivity, plus you marry that with your critical thinking and insights and your influence, your business partnering, this is gonna be the best decade of your career. basically you just need to upscale yourself how to use these skills and this technology. what we are getting every day. It's it's a huge improvement just to be able to catch up with that, stay on top of Giac: Yeah. Arif Ender: it, try to utilize effectively, but marry that with those skill set, better strategic communication, critical thinking, analytical. trying to grasp filter sometimes or evaluate what AI gives to you. Don't take it right away granted. I think right now it's so obvious, you know, AI generated stuff. I I think we can see it from miles away. Make it basically evolve it in a way, with with your hands, that's where it becomes Giac: Mm-hmm. Arif Ender: really, really valuable. Giac: Excellent. Arif Ender: first of all, we we are trying to like really boost our processes at my current company with AI. obviously we are I I'm in cybersecurity and we are one of the frontiers in my industry as a company, utilizing AI and also securing AI. So it's it's quite an interesting phase that we are in as a company. Giac: Mm-hmm. Arif Ender: so that's why it's it's always in in our daily life, you know, how are we going to use AI or AI, AI, AI, left, right, center. So the only problem that I have because we are also we know the risks of this emerging technology, and having kind of a zero trust mentality, that's not pretty normal, I would say, in cybersecurity. basically the tools that we take on board is a bit slow. But we are getting there. So I think the last time we spoke, Jack, about it, like things changed. So now we have agent designers and stuff for us. So we all automatically started with a roadmap. So how can we utilize Agentic AI in compensation? So we laid out like a couple of projects, you know, did did some sort of an impact analysis, effort impact analysis, let's say, you know, like how much will it take, what do we need? And how do we get there? But we immediately started like with two agents. Now we are just trying to kind of finalize. We are actually stress testing the prompts about one is offer agent. we just want to automate the the offer proposal proposals. The other one is we are utilizing a couple of gems. They are not maybe agentic, but you know, for job architecture, job matching, One is data visualization for actual business partnering. when we are doing like trainings Giac: Okay. Hmm. Arif Ender: or kind of like a briefing during the pay planning Giac: Mm-hmm. Arif Ender: cycles we have twice a year. We are utilizing the data visualization, we just kind of like feeding all the all the pay planning data and utilize Gemini to create all the data visualizations and decks for us. So it takes almost like three, four days. for each business partner, for comp comp business partner to do that, it becomes a couple of hours max, j it even just basically iterating correct, you know, auditing and stuff. So it's it's pretty cool. On the personal note, I'm a proud Cloud Max subscriber Giac: Nice. Arif Ender: although it's not cheap, but it pays off. I think it's probably this year's best investment. From a skills development perspective, I'm learning a lot while I'm wipe coding during evenings and weekends. So I'm working with some ex colleagues, some friends for a couple of projects, let's say we call it, you know, side hustle. let's see how they they are gonna go. Mostly again some comp products but with a different twist. so you'll hear hopefully you'll hear more about them soon. Giac: I'm yeah, very intrigued for sure. Arif Ender: Yeah. Giac: and and it's not that someone needs to start from building into a compensation platforms, right? I think you can start by like you can really ask Claude or Codex to write scripts that you run locally in your terminal to do some of the grunt work, to automate some of the you know, steps inside a workflow. Be it salary band design, be it pay reviews, that kind of that task that no one wants to do, that AI is good at, there's no spill of information, stays in your computer. And that's a good place where to start, right? We we we've designed several of these scripts, and the the the community goals we have in range are precisely you know, to help people start with simple things that make a huge difference. But it's also great to hear that you're doing bigger and thinking about actual, you know, like end to end applications that y eventually once you get to that experience level, you can ship across your company. Arif Ender: it was also pretty interesting because I stress tested. I'm more cloud user, but I I try to utilize almost all of them. For instance, for deep research and like quick search, I also use perplexity. And then for probably Giac: Mm-hmm. Yeah. Arif Ender: generally more text generation, I also use Gemini or and Chat GPT kind of. I I mix and match, but my eighty percent I am on Cloud. And what was fascinating is from very early career I I built a modeling, a certain modeling that we we actually also asked an external vendor to help us, which charged us almost eight thousand dollars or something and shipped within four or five Giac: Wow. Arif Ender: weeks. I uploaded that model to Cloud. Can you do this web based? Fifteen minutes, it was done. Man, I I I said, Okay, yeah, this this is gonna be game Giac: Game changer. Thanks for coming to the podcast, Arif, and for sharing your knowledge and experience. If you had one piece of advice Arif Ender: Thanks for Giac: for a compensation slash total rewards leader that is feeling behind the AI curve and wants to invest their time wisely in the next couple of weeks, what would that be? Arif Ender: Well, you know, it doesn't have to be in the company. Just think about it, like j even generalize it. What what what is your number one, number two, you know, let's say top three problem that you face as a challenge, as a stress factor in your current job or even maybe in your past roles. And really sign up. maybe at least for one month, you know, just pay for a for a good Cloud or ChatGPT. And start building it and reiterate. Use a couple of best practice prompts. That that actually is is a really good one that I can recommend. You know, asking asking also AI to challenge you and asking always for no assumptions, ask clarifying questions. Do not take shortcuts. I I use these a lot and it really changes how I interact with AI as well. Because you also learn as you go, because it it asks you something you you are never exposed. You know, should I do Giac: Mm-hmm. Arif Ender: this in terms of security? You know, should I add this feature? I never thought of that. Yeah, it makes sense. And build it. I I think it sparks some ideas or Maybe it just directly gives a product to you right away that you can Giac: Right. Arif Ender: you can utilize. It's a thought partner. Giac Soliman: That's a wrap. If this was useful, follow the show and send it to the one other comperson you know who needs it. Thanks for listening and I'll see you in the next one.