Latest / Range Podcast / Build the thing nobody asked for

Transcript

Build the thing nobody asked for · Range Podcast

• Giac Soliman: Welcome to Range. I'm Giac Soliman and this is the podcast for comp and 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. Good morning everyone and welcome to another episode of The Range Podcast. I'm here today with Martin Smith. Martin is a former Toll Rewards leader. He spent 25 years across large multinationals, pharma, oil and gas, you name it. And if you don't know Martin, well you probably know him from LinkedIn because he has been posting some great tools and experiments. He's been building out of the open For quite a few years. Welcome welcome to the podcast, Martin. Martin Smit: Thank you. Thank you for having me. It's great to be on the. Giac Soliman: So let's dive a little bit into your experience with AI when it all started. I'm interested in just even before AI, I understand that you have a very particular background, something that you do not typically expect in Total Rewards leaders. Is that right? Martin Smit: Yeah, it's interesting indeed. So I call myself more techie and some of the trackie, although I love Star Trek, but I started actually coding in 1984. This was before even PCs existed. So fascinated since then. So that is 42 years ago. I started programming. Then always been fascinated with technology. almost studied artificial intelligence in 1996 when it was at my university in the Netherlands, but I learned it is 80 % statistics. So I opted out for the business studies next door, which would bring me to all kinds of exotic locations. So that's how I rolled up basically in rolled into business and then started working in HR. Then in about 2017, I... heard about AI at the time and I thought, okay, let me see how much it has evolved in 20 years. I haven't really looked into that. And I was absolutely amazed in 2017 already what it be able to do. So I did a study at MIT, a lot of reading, actually wrote some papers, very short. One paper was that it would be nice if we're able to use human language, like chat, to communicate with AI. It wasn't possible then. Giac Soliman: What a coincidence. Martin Smit: Yeah, that was a bit early, but unfortunately no monetization on my part. But, you know, very interested, but at 2017 AI was very complicated. It was not easy for people to do. You needed big systems and there were only actually, you know, a dozen influencers on this topic and maybe a few hundred companies. And I even inventorized all of them on an old website. And I think 90 % of the companies don't exist anymore. So I didn't do much with AI. But then in 2022, 23, maybe it's the same like you or your fellow readers, I heard about ChatGDP. So I thought I'm gonna explore that a little bit. So logging in and I thought, oh my God, this is absolutely amazing what the tool could do. That was 3.5 or earlier. So basically that opened my eyes. And from that moment, I've been more or less fully deep diving in mostly weekends and evenings, trying to build things, learn things. and the last few years also more in my work. So that's a bit of my background. Giac Soliman: So I've I've got to ask you this, Martin. Do you recall what was the first thing you typed into Chat GPT and what you got back? Martin Smit: I asked it to discuss with itself about, I wanted to see how a discussion would evolve. Like I said, discuss with yourself, I don't know the meaning of life or something like that and do a role play and was fascinated to see the discussion. I think that was one of the first questions asked. And again, if you look back a few years ago, like most people, it was absolutely amazing, right? Now, if you now look back, And my God, this was entirely basic, right? And that tells a little bit about where we're going to go in the next few years as Giac Soliman: Excellent. I I've asked a few people the same question I get. Some some people asked questions about themselves, like they would ask who's Martin Smith and they would get an answer. Now we all know that Chat GPT's like knowledge base had a hard cap at I was twenty some twenty twenty two or something and then they moved it to twenty twenty four, like it kept on moving. So the answers would wouldn't always be relevant. And other people asked them, you know, draft a fully fledged total reward strategy that I can adopt tomorrow. And yes, at the time to your point, you would be like amazed. Like, look at the amount of text, and it all makes sense. Now, if we read something like that, we would probably know, this is AI generated, and we would discount the value of it. But back in the day, it was definitely something alluring. Now you said in your intro that you, thanks to your work, you traveled to exotic places. Now, if you had to choose one place like one last day in my in my life, that's where I want to spend it? Martin Smit: Well, that's a very human question and the HWAI, would say, that is very Giac Soliman: Deep. Martin Smit: interesting question. Well, it's not an exotic location, but it would be Switzerland. So I moved to, I've lived in the Middle East, I've lived in Asia, I've lived in certain parts of Europe, traveled to probably all continents. But I feel really at home here in Switzerland. It's beautiful environment and... That is at the moment where I would then would breed my last bread, hopefully in many decades from now. Giac Soliman: let's go back to our AI discussion and Martin Smit: Sure. Giac Soliman: I want to name the thing that I don't hear enough people saying out loud and I'm very interested on what is gonna come out of you, Martin. What's the biggest mistake you see people making with AI in total wars right now? Martin Smit: Well, again, mistake is a loaded word, right? Because it's not necessarily something as right and wrong always. I think generally the biggest mistake would be not to make any mistakes. I think what I mean to say with that is people not trying to use AI, right? Because I can talk a bit about the barriers, but, and I still see that happening a lot. I still people seeing not even trying it or experimenting with it. And I think that is, itself a mistake, because you're not really trying to learn a new technology that could be completely disruptive to society, to your job or to your own situation. So I think, I think that is one thing that I generally see. That's not constrained to total rewards. And we can talk a little bit in a second about some of those barriers that I still see after several years of talking about it. Total rewards, you specifically asked me about. I think, and again, I'm just thinking out loud, too often people quickly jump in using AI for the things that they spend time on. Like for example, compensation benchmarking. That's often the first discussion. can we use AI to do pay benchmarking? I said, sure you can. I I can build a tool, you know, maybe in the weekends that would go to competitors' websites, scrapes data, consolidates it. Well, and then what? Right? And then you need to do validation. How do you know that? the jobs aligned with each other and where's the value coming from? So I think in that space, for example, I think there's excellent tools out there from vendors. And yes, it may cost a bit more and do much yourself, but I think you get higher quality, you know, get what you need. And what I mean to say with that is that I think in rewards and many other spaces, we need to start looking at AI. What are other areas where we can use AI to generate more value? And of course you can generate some efficiencies, but If everybody's focusing on using AI for pay benchmarking or everybody's using it for the equity analysis, we're going to have a thousand different ways of doing things, which is not necessarily bad, but not efficient. And we're not touching all of the other areas which can add more value. So that is generally what I would say. Giac Soliman: I'd love to follow up on that 'cause this is getting very interesting. So the first like elephant in the room and I I'm feeling this every day. Like I'm realizing more and more that just because I can build stuff doesn't mean that I should. Martin Smit: Yes. Giac Soliman: I thought it would save me time, to your point, it's all based on efficiency, but then I realized, hang on, like I was just amazed by the fact that I can build stuff, but does it actually work? Does it serve a purpose? have you developed a a framework, even if it's just mental like living in your mind, to decide when it's worth investing your time and energy to build something and when it might not be worth it? Martin Smit: Well, I think it's a bit of a schizophrenic thought, but again, on one hand, I think trying out stuff and I'll talk about some of those crazy ideas as well. Even if it makes in the end no sense, it does make sense because again, you're learning, you're learning a skill, you're learning how to prompt, you're learning how to structure data, you know what the system can do. On the other hand, of course, if you... Every time build something and you're not doing anything with it, you're also not coming at a good end, right? And I was in that situation or probably maybe even still am, but by every time when I had an idea over the last year, especially when cloud code came up, like, I wonder if I can build it. So I came back from Paris and beautiful museums. think I wrote it on my LinkedIn. And I was just saying to myself, I wonder if AI can build me a site on art that doesn't exist. That's why I just go to cloud code. said, bring me in concept and I ask them to go into plan mode, right? Develop and plan for a site that doesn't exist. And before you know it, I spent two weeks basically in the evenings working on an art site. And yes, it has 4,000 museums, 200,000 art pieces and stuff that I think other sites don't have. Probably not many people are using the site. So you could say in one hand, it's a waste of my time. On the other hand, when I look back at... that site or a gardening website or all the other nonsense that I made. Every time when I do that, I so this is how I need to structure data. Aha, this is how I need to instruct systems to basically review it. I can share a bit my views about what I think is harming most organizations. I see AI, let's say it's a new tool, like a new play tool and we're kids as humans. And we need to play with this tool to understand that. And kids also go playing outside and make sand castles that will be destroyed after that. But they're learning a skill and they're learning to conceptualize what they're doing. And I think we need Giac Soliman: Mm-hmm. Martin Smit: as humans to allow ourselves and our colleagues to do that. if I can just, another thought which I often talk about and I don't see this really moving is I think a lot of organizations and maybe even the larger ones are putting unnecessary barriers up to allow people such as you and me and others to experiment with that. And I was recently at a large company, we had a lunch conversation and they said, our IT department doesn't allow us any tools other than the simple chat Giac Soliman: Mm-hmm. Martin Smit: GDP. And somebody else said, the legal department doesn't allow us to use anything. So. They're basically living in a situation of two years ago. And of course then they're not getting the skills. So what I always say to organizations is give your people the tools, but in a safe environment, right? So if they're messing it up, not basically the data goes outside, right? And IT can handle that and give some guardrails to people what they should and should not be doing. And when we're talking about reward, Too often it is about, well, we're talking about reward. We want to use it for people decisions and saying, well, you just need to be careful. You can do a lot of stuff with AI, developing job architectures, know, speeding up mobility processes, reviewing, I don't know, executive compensation data without necessarily having to input employee data, without having to make decisions. So. If organizations just give the tools, make sure it's sandboxed and guardrailed and give some guidance, I think they'll be developing the skills that they need. And then they will play. They will make nonsense websites, But they're developing a skill and that I think will be very valuable. Giac Soliman: Yeah. It is true that especially multinational firms like Whenever a new capability becomes available, you know, it needs to go through IT procurement. So understandably, it takes time to roll out new technology. But equally to your point, you don't wanna be prohibiting the use of these tools if there's a better way to use them safely while you figure out let's say your A AI governance strategy. I heard of people that have been told no and they've asked for a sandbox environment to play And to test new technologies where with with dummy data, there's really very little risk. But I think it's becoming part of an organization employee value proposition. Like those that have been able to move quickly with AI and are encouraging safe and effective use of AI are might be seeing I have a feeling that this is impacting and will impact retention as well, because more and more leaders are feeling either behind or stuck with AI. And I'm interested in what your advice would be if you're a toll rewards leader in a company that's so large and moves slowly that is promising you AI tools, the ones you want, in a year time, and by then the technology might already have changed dramatically. six months just we were all banging the drums on Chat GPT six months ago and now everyone you know hyped a cloud code. Like what would you what would be your advice for people and you've been in that hot seat as well, Martin Smit: I'm Giac Soliman: right? What what would be your advice to help them get started? without necessarily having to wait a year for their company to figure out what their AI governance strategy is, what the right tools are, et cetera. Martin Smit: Yeah, so first of all, it goes very fast. coming back to that point, right? So a year ago, I think Cloud Code wasn't very well known. I think I only started in October. I'm only using Cloud Code. I mean, I hardly use Jet GPT. I do not know what the AI world will look like in six to 12 months. But to be honest, I cannot keep up anymore. Every two or three days now, and I see something and I said, oh my God, right? Impossible. Three months ago. is now possible or faster. it is really, it keeps going on and every time also in the media, everybody says it's plateauing. No, it's not plateauing. We keep seeing it going up and that's partly because we're having millions of people using AI thinking of new ways. And there's just one person that comes up with a great idea, puts it on GitHub and a million people can use it an hour later. And I think that's really gonna continue over the next few years. So think coming back to your question, right, because I was diverting a bit. So I think first of all, have your IT, legal and HR, not necessarily toe towards, sit in a room and basically say, how can we give, let's say the people that want to experiment and needed safe tools in a safe environment, right? And some guardrails. think that you just need to get over that. Otherwise what's gonna happen? People are gonna try it at home. right with their own tools, not company data, but you know, that will be a missed opportunity. The second thing, if I'm thinking about rewards, but I personally see rewards as a test case, right? I've also been an HR business partner and talent and other functions. I just happen to know a little bit more about reward, but I would recommend functions, not to think so much about what can we do with AI, but it's more to think through what do they want. And to come back, was recently in another conversation with an organization and one of the leaders, I think it was Benefits, me a good question and he said, well, how can we level our benefits with the help of AI? And my response was, well, you need to define what that means, right? Because if I always say AI can do everything, so what do you want? What works for your organization? And... And I would really encourage people, especially in leadership positions, to take a seat back. And I, what I prefer often as a Greenfield is ignore all the constraints that you have, right? And all of the stuff and all of your governance models in the ideal situation, what do you want as an organization? Your strategy, then what do you want to do as a function? What do you want to do as total rewards? And then figure out how can AI play a meaningful role in that? And I think if we do it that way, we're gonna move out of the efficiency space, right? Or by, this can be done faster and can be done cheaper. Yes. And we can start moving to the value space. How can we use AI to generate more value and to become more competitive advantage? And I would encourage that. But develop skills, take a step back about strategy and do that ideally in parallel. That will be my high level suggestions. But caveat is it all depends. right on your organization, on your maturity, on the systems you've got access to, on the budgets. another comment I often hear or question is, I was in a forum with a lot of small biotech companies and one small biotech was asking, well, are we small biotechs not an disadvantage in using AI compared to the large MNCs or in the room? And my answer is, I think it's the opposite. said, actually, you don't need a lot of budget. If you look at what I spent, I spent $500 per month on everything. And that doesn't mean I can run a pharma company, but I can do a tremendous amount of stuff and they don't have all of the governance models and the problems. So I think there's a lot of opportunity and companies just need to start. Giac Soliman: Excellent. And that five hundred dollars a month, it might sound a trivial amount to some. It might sound like a big amount for those that don't really understand what the potential, what the AI potential is. Martin Smit: Yes. Giac Soliman: And I think when it comes to AI, it really helps when we like showcase just an example the things you can build with AI that are to your point must be linked to what you want, right? Martin Smit: Yes. Giac Soliman: Not it just because you have this new capability, Everyone's asking, what are use cases? How can we reduce churn? How can we become more efficient? But to your point, it's really about what you want. Martin Smit: I'll give you everything as possible, right? As long as you define it and think through it. Giac Soliman: I was wondering if that expense you had also includes the Claude Claude API 'cause I find myself like I I might spend like five pence you know or cents calling Claude API with Sonnet or Just curious if you're using that at all. Martin Smit: No, no, I don't use. So what I use on my tech stack is I use cloud code and then all the API calls I mainly make either to Gemini, right? Google cloud and also I use open AI and the only reason why I actually use open AI is I get like 10 million tokens for free per day because I am a subscriber for a long period. I do have API access to cloud to Mistral to some other tools. Sometimes I'm comparing them a little bit, but again, what I do is saying, well, where do I get most value from in terms of cost? And that is where I focus on and Gemini 2.5 flash and 3.5 flash. I think give great value for doing a lot of stuff. mean if the same is what I noticed with with cloud codes, I was again a mark subscriber and I got this email. I think over the Christmas holiday thing or because it's Christmas you're going to get double or triple the amount of of tokens right now. Of course there goes your Christmas holiday, right? Because you try you try to bridge the limits, but I've not really experienced any any any hits. hitting my limits. did once have a security breach whereby apparently somebody, I'm not sure if they took my API keys, but they were able to generate suddenly a lot of images with one of my API keys. So Google then basically luckily stepped in, that thing because you need to build that in. Otherwise you're going to get a $15,000 bill in the morning. So I think I had it at a hundred or $200. And then I spent two days of auditing everything was there was and it turned out actually it wasn't my mistake. It was something on the Google site, but $100 is not worth fighting over. But a big learning from that is before we talk about executive comp. And again, every week I'm learning something right? But a big learning from that is is you really if you're going to do vibe coding and especially at home and it's your own money, you're going to have to put those things in place. You're going to have to put caps in place. You're going to have to put kill switches in place. If not. It could be a morning you're waking up to a $10,000 bill. And then that Giac Soliman: Yeah. Martin Smit: is really very, important for people to consider. Giac Soliman: And I I think what people misunderstand most of the time is that if you get a ten thousand dollar bill is not necessarily due to your clot setup, is due to your own behaviors. Like you have to learn how to use these tools in a way that it doesn't become unrealistically and unbearably expensive for you, right? Martin Smit: mean, one night I had it embed stuff. I wasn't really clear. Embedding means it's processing text into something. I wanted to have, I have a database in it with 10,000 companies in there. That's my executive comp project. I tried it for fun. So I had 10,000 companies in there, the largest world. Then I asked it again, I'm simplifying it. said, for every of those 10,000 download the annual report in the proxies. It did it. That doesn't cost a lot of money. And then I said something like embedded, right? Now without being very clear, I've got it in auto mode. So in the morning, then I look at it and say, what, what did I spend? That's another thing. What did I spend the hundreds? Oh yeah, we've embedded 600,000 fragments of that. And I said, Oh, said, okay, I should have been Giac Soliman: Yikes. Martin Smit: clear, but I mean, you know, again, this is for me acceptable, right? I mean, I had to probably do it anyway. $50 is a good learning, but if you're not, if you're really not know what you're doing, and you don't have these gaps, things can get very ugly. But coming back to the exact comp, I most of my experiments or things start with me wondering, oh, would this be possible? And that's what I often try to do. I see something early in the week, again, somebody posted, they're able to generate a voice call center. I mean, me being me, ooh. I wonder if I can do that, right? Because I don't want to pick up my phone and answer questions. So yes, there is a GitHub library that in an hour you can actually Giac Soliman: Love that. Martin Smit: connect to in VoIP and actually you can have for $3 per month your entire computer automated call center setup. Now, I mean, of course that's going to be completely disruptive, but that is every time when I see something I wonder like, that possible? And I think half a year ago, I think it was at a meeting or conference. I just wondered, could I basically do this? So, and then some of the things I try and then after one or two weeks of, you know, trying, just say, okay, forget about it. It goes in a dormant phase. And some of those projects saying, this is interesting, right? For me, it's personally or for users. And one of them was executive composition. And the beauty of that is that it's public information, right? It's, it's again, reports. Giac Soliman: Mm-hmm. Yeah. Martin Smit: So yeah, I spent some time on that as well. Giac Soliman: Understand this is work in progress for you 'cause there's always way to improve it, but to help to help, you know, people are listening in visualize what you've built and what is possible with AI, would you be willing to share Martin Smit: I can. Giac Soliman: free maybe some screenshots or maybe just the Martin Smit: Yeah, I can. Giac Soliman: the interface of of what that looks like? Martin Smit: Yeah, I can. I can share. And again, it's still work in progress and it all comes down to when I've got time, right? And it's really coming down to I need to make time for myself, but this is this is even an older version. right? So I've got I've got about I think 10,000 companies in this list. So I basically have literally the world's largest 10,000. Then basically had, when I say I, it's basically cloud code doing the work for me, right? I just telling it what to do and what to do. Yeah, I'm just, I'm not doing the work Giac Soliman: Yeah. You're the orchestrator. Orchestra yeah. Martin Smit: other than telling it and saying this is wrong and that should be wrong. But for Giac Soliman: Amen. Martin Smit: all of those 10,000 companies, it has downloaded the annual reports, the proxy reports, right? Now you let it run overnight for example, and then it sounds simple, right? But of course it will cost a bit more work. And that is where I spend a lot of time and saying, okay, if we have that, how would it be able to extract data? And then you need to think about how, how would data look meaningful to your end customers? Right? So an exact comp professional or an total work professional, what are they looking for? What is possible? How would you extract it? And that is, would say where the combination of domain expertise as well as customer focus Giac Soliman: Mm-hmm. Martin Smit: and tech will come in, but you want to have stuff on governance. You want to have extracts on compensation philosophy. program design for every executive. It would be changing control terms, comp tables, NEO, KPIs, right, which is useful as well, and payment for performance. Giac Soliman: There's so much. Martin Smit: I mean, it extracts the peer groups from the companies and you can compare Giac Soliman: Yeah. Martin Smit: that. And then once you have that, of course, you can build a tremendous amount of tools. So when hopefully in the next Giac Soliman: Wow. Martin Smit: few months I get this working, Let's assume I've got 10,000 companies in there. I've got for the last four years, all of their data completely extracted, what we call embedded. Now the amount of analytics you can do is mind boggling. And I would say that is in most tools these days, not possible. So this is an example where I really, I'm quite passionate about. And hopefully we're going to see that a bit more evolve. And it's really. I mean, I think this is version 749 or something by that time, right? So every time you do something and you fix one thing, you break another. So that's quite a good learning. Giac Soliman: this hits home. I I started my career in executive compensation as a consultant. So I was literally paid to read all of these reports and distill findings from them in a way that we sort of can statistics on top of it. it's it's mind boggling the amount of information and insights you've been able to to to summarize in here. I we're We're looking at Nvidia or NVIDIA, disclosure levels are not always the same and you also encounter differences across between different countries, right? How how would you handle something like that? Martin Smit: Well, that is indeed where a lot of trial and error comes into, right? Because I mean, we're speaking to total watch professionals. US is relatively structured. You've got your 14-8 reports there, which are basically even HTML. Now the rest of the world, it is not HTML. It's all PDFs, right? In Giac Soliman: Yeah. Martin Smit: different formats. And some have a 400 page document, some 100, some combined with any reports. Some of them nice graphs. China does it differently than Japan, than Korea, than Germany. So the trick is, and this is where lot of the effort comes in, is saying, okay, so what would be a global view that a professional or a function would wanna look into? And how do you make sure you extract it? And of course you accept that not all data is always there, right? So in the US there may be data on race, for example, in the reports and it also extracts by the way from the annual reports like how many employees there had, what the makeup is. Other parts of the world don't have that. And then you need to make a choice as a designer of a system. Okay, if only a certain part of the world has it, am I gonna show it or not show it, right? And show it not available. And this is again, where it's a bit of thinking out. And at the moment, my view is, well, I'd rather have more data because it's always easy to say, don't show this or hide this. But in essence, what I'm trying to build in here is extract as much of possible as data. and then we can always scale it down. But again, that's a lot of work. the challenges is mainly non US by the way, right? Because that is where you Giac Soliman: Mm-hmm. Yeah. Martin Smit: have graphs and PDFs. How do you? I find it important, especially with this, that if people look at the number that they can see where this is number come from. Not like, I need to trust this, right? What it should show is it should show actually the page and actually say, this is the page where the information comes from, right? And Giac Soliman: Yeah. Martin Smit: that stuff is not so easy, but I'm confident that I'll crack it. But this again, this started as a thought experiment. And by now I've got quite 10,000 companies, three years of reports. not just proxy, but also annual reports. I could turn this around and say, I want to analyze ESG or even financial analysis. It's the same concept. And therefore I often say, total rewards is an excellent playing ground. But the things you learn in total rewards, you can apply that to other fields. It's basically, how do you look at very complicated data, turn it into structure and turn it into something that you can make for decision. Now, another thing, and just you brought me up to that thought is all of this started with thinking is when I was working on my own bots and agents, I often had difficulty in getting data, right? For example, I wanted to have a chat bot that does ethics calculations, right? Well, Giac Soliman: Mm-hmm. Martin Smit: if you're not configuring that, it's gonna probably go on Google somewhere or on Bing, find something and bring it back or a share price. Sometimes Giac Soliman: Right. Martin Smit: it works, sometimes it breaks, sometimes it's outdated, sometimes not. So it's really that made me think like if we as total reward professionals and maybe finance professionals want to start using more chatbots to do stuff, you don't want to spend like I did half a year ago, two days on trying to configure the tool, how to get an FX rate. So again, and not a joke, but as an idea started well. Would it not be possible to create something whereby companies can basically get all of this data through and simple API slash MSP interface. And you can see right here, I've got now I built this. It's not officially live yet. It has data on 220 countries updated daily, 400 fields, 410,000 records. And it goes all the way from family benefits, healthcare statistics, income taxation and all of All of this comes from very viable data. Like, okay, this comes from the ILO at this date. And then companies can very easily integrate it in their tools. So again, this is also completely built with cloud code. And I think if, if I'm able to, let's say, make this workable, then a lot of companies or people like you don't have to worry about this and can say, I want to work on something with my company. I want to plug in whatever benefits data or severance statistics. I'll just connect with this. And this is just a second idea. And then another one, ideally we think out one other thing, which you often asked. And again, this is all me often building in the weekends, pay equity analysis. So I'm of course being involved in my profession in that area. And I would say the MNCs from what I know are relatively advanced in space. Most of them are working on this for years. There's quite some mature and solid solutions I would say at the global scale, but you're gonna have a hundred thousand or 200,000 companies in the EU that need to do pay equity analysis in the next 12 to 24 months. And they're gonna be in two countries or four countries. Tools are complicated, laws are complicated. So, and there's a lot of concern on data privacy. So in one weekend, I said to myself, I wonder if I can build a tool that you can run on your own computer. This is not a website, but by you can upload data and basically it will more or less do everything for you. It will, you can also upload additional data like a job catalog if you have it or your own leveling framework. And it will analyze which countries are it will say this is the law. Do you understand that? Again, this is all just, I'm just going to walk you through a little bit of what it can do. It analyzes everything. does mitigation. And what I liked about this one is, and this is all dummy data, by the way, this is all done on your PC. So if I'm able to finish this and you would run it or anybody. There's no single data would leave your PC. It's just run on your PC. Analysis Giac Soliman: Right. Martin Smit: and also build with cloud. Just an example. Giac Soliman: Yeah. Look, this you said there at the beginning of of of the podcast that, you know, y think about what you want to do, not just how can AI you know, make me more productive or save me time. and I think these are great examples of it. I think it's part of someone's imagination, right? You could you you linked your domain expertise to your imagination. You found out that streamlining executive compensation, benchmarking and decisions it's something you wanted to focus on, but also pay equity, 'cause to your point, there will be thousands and thousands of companies that didn't have to think about this until June twenty sixth. And suddenly now they have to. I want to end up with a sort of forward looking note. Martin Smit: Okay. Giac Soliman: since it's about imagination and it's about dreaming a bit, I want to help other listeners dream as well. And I want to ask you if you could wave a wand and fix one thing in compensation or total rewards. Forever. Something that really stuck with you while you were in the job no matter how impossible it feels what would that what's that headache of the toll rewards leader life that you would love to make disappear Martin Smit: Well, first I'm going to say the thing which I dread the most, but that's not my real answer. I think the annual performance review or compensation review, I always say that's process that needs to be killed. I mean, I think it's like so much energy is spent in organizations and thousands of managers and employees on it, but that is not my official answer. But I think I just, I don't understand why most companies still spend so much time on that. we need to step out of the silos. So what's happening is you've got your compensation team, you've got your benefits team, you've got your equity team, you've got your mobility team, and then you've got your talent team and you've got your talent acquisition team, Historically, it was normal that because work needed to be divided up, everybody's working in the little silos. With the help of AI, you can do things completely differently. And I think what I would hope is that organizations take a step back and say, what are we able to achieve with the help of AI that may require a bit of an organizational change, but changes to a work and we get much more out. And as an example, and I've built a proof of concept, so I always know it's possible. I'm convinced at one point there will be something like an AI whereby you can say, this is my strategic intent. And this is my restrictions, right? My resource issues are my budget constraints. And it's going to come back with everything, Organizational model, the jobs being leveled, the actual cost implications of that, where you may need putting those locations, which outsource vendors you might need to do, how it's going to integrate. And AI could do that, right? It's not a question. If it has the data and the willingness, you can do that. And what I would hope is that at one point, Leaders take a step back and think the way how we've run organizations, way how we've done total rewards or compensation, does that still make sense in this new way of world of working? And I think that is a good thing to look into. 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.