"I Hate AI": Karim Gharsallah on Why a Tool Is Not a Strategy

Show notes

In this episode of the Leaders in Talent podcast, Adriaan Kolff sits down with Karim Gharsallah, Head of Talent Acquisition at BDR Thermea, the 7,000-person European manufacturer of boilers and heat pumps. Karim opened their prep call with three words: "Adriaan, I hate AI."

The nuance is what makes this episode. Karim does not hate the technology, he hates what it is doing to how we communicate: every LinkedIn post with the same structure, personality fading out, and companies believing a tool will fix problems they have not defined. His line for it: a tool is not a strategy. Meanwhile he uses Claude every day, for the deep work of building the business case for a centralized TA function and for structured interview guides, and BDR Thermea's policy has just moved from "no AI" to public-information-only.

He walks through the Recruitment Blueprint he is building: why companies that start by hiring recruiters end up with inconsistent processes, unclear ownership and agencies everywhere; taking ten steps back to define what recruitment should be; governance and hiring guides first; a baseline across seven countries; intelligent alignment instead of centralization for its own sake; the Ulrich model; and a flexible mix of roughly 60-70% in-house with RPOs and agencies for the rest.

Karim and Adriaan also get into the move from scale-ups (VanMoof, and Recruitee where he was Global Head of TA) to a 7,000-person corporate: decision speed, pacing yourself, and why every company thinks it is uniquely complicated while having exactly the same hiring problems.

Timecodes

01:09 Welcome and Karim's background

02:54 "Adriaan, I hate AI": the nuance behind it

04:39 Where Karim actually uses AI, and why he switched to Claude

05:22 BDR Thermea's AI policy: from "no AI" to public-information-only

06:38 AI use cases for a head of TA

07:46 Karim turns the table: AI fatigue

09:41 How Matchr is doubling down on AI

10:40 Rebuilding a website with AI instead of a 20,000 euro project

12:49 The Recruitment Blueprint: take ten steps back

14:21 BDR Thermea: 7,000 people, decentralized by acquisition

16:50 Centralize or not: intelligent alignment

18:30 The Ulrich model and building the baseline

20:57 "Every company thinks they're uniquely complicated"

21:49 From scale-up to corporate: the speed of decisions

23:24 AI principles before AI tools

25:26 Where to follow Karim

___________________________

Connect with us on LinkedIn: https://www.linkedin.com/company/matchr/

Get in touch with us: https://www.matchr.io/who-we-are/contact/

___________________________

Connect with Karim Gharsallah: https://www.linkedin.com/in/karimgharsallah/

Connect with Adriaan Kolff: https://www.linkedin.com/in/adriaankolff/

___________________________

RSS feed: https://media.rss.com/leaders-in-talent/feed.xml

Transcription

[00:01:09] Adriaan: All right, ladies and gentlemen, welcome to another episode of the Leaders in Talent podcast. Today my guest is Karim Gharsallah, the current Head of TA at BDR Thermea. Karim, welcome to the podcast. Can you maybe give our audience a little bit more about your background and what you do at BDR Thermea?

[00:01:30] Karim: Yeah, sure. Thanks, Adriaan. So I'm Karim. I am currently the Head of TA at BDR Thermea. We're a manufacturer of boilers and heat pumps, focused on the European region. I love TA, I think that kind of summarizes what I love to do. I've been in the field for 10 years. The energy that you get from finding the right talent, but also contributing to the strategic direction of a company, is just really fun. On the personal side, I'm married since last year. I think there's a picture in my background, just above my hoverboard there.

[00:02:05] Adriaan: I see it. I was looking at the hoverboard. There's a lot going on in the background.

[00:02:08] Karim: And you can see a whiteboard. In the corner you see Operation Ido. I used this whiteboard to organize the wedding that we had last year. So just a little sneak peek. Anyhow, that's a bit about me.

[00:02:21] Adriaan: Great. Awesome. Hey Karim, many people say they never expected to end up in recruitment, or fell into recruitment. I don't think there are many people that work in recruitment and thought when they were younger, "I'm going to work in recruitment and selling boilers and heat pumps."

[00:02:36] Karim: Yeah, I think I'm one of the few who knew that I wanted to start my career in recruitment. It was the second year of my bachelor's, I was doing an internship, and I was like, "This looks like a fun field to be in." I knew what I wanted to do, and yeah, there you go.

[00:02:54] Adriaan: There we are. Awesome. When we had our prep call before the recording of the podcast, you told me something interesting, and I really want to kick off with that. You said, "Adriaan, I hate AI." I was like, "Wow. Okay. Tell me a little bit more about that." And we almost recorded the full-on podcast before we even recorded the podcast. So let's kick right in. Why did you say that?

[00:03:15] Karim: Okay, so I think there's some nuance behind it. In general, I don't hate AI, but I do hate AI for a lot of things that it's being used for. One is that there is a certain generic feeling about the way that we are communicating, sometimes in my own company, but also sometimes on LinkedIn. Every post has exactly the same structure, and it feels like that personality is gone or starting to fade out, and I think that is a shame. The second layer is that people talk about how AI is going to take all our jobs, and maybe it will, I don't know. But to me, a tool is not a strategy, meaning that you can't just go blindly implement AI tools thinking it will solve the issue or the challenge that you have in an organization. So on one end it's a personal thing, I just hate the generic tone of voice that AI gives us, and on the other hand it's thinking that it will fix all of our issues, or every company's issues.

[00:04:27] Adriaan: Yeah, interesting that you say that, in terms of LinkedIn posts where everything feels like it's been written by AI. So tell me, where do you use it, or what's your philosophy around AI?

[00:04:39] Karim: I use it for the work that I do. I use Claude. I switched from ChatGPT to Claude, and I use it for deep concentration or deep self-work, where I want to have a conversation with, in this case, an AI to layer out the work that I'm currently working on. Whether that's a recruitment strategy or something else, it adds another dimension and another deeper layer to what I'm trying to build. And I think that is wonderful, but I'm in control of what it is that I am prompting and what it is that I'm asking. Every question that I ask is another layer that you peel off to get where you want to go.

[00:05:22] Adriaan: Yep. And what's the general strategy within BDR Thermea with regards to AI? Is it fully adaptive, does everyone have access, or is it "let's first make sure that everything is compliant"? Where do you guys sit?

[00:05:35] Karim: Our initial policy was no AI, and that was recently switched to: you can use AI as long as you only use public information around the company. We are running some pilots when it comes to Copilot, but a very limited amount of people have access to that, and Copilot is fine, but it has some limitations. So everybody uses their own ChatGPT or Claude.

[00:06:01] Adriaan: So that is allowed. That's interesting.

[00:06:03] Karim: That is allowed as long as it's public information. Yeah. So we're not that far yet.

[00:06:09] Adriaan: If you look at your recruitment process, is there any tooling that you've implemented, or anything that you're considering?

[00:06:16] Karim: Not yet. We're in a hiring freeze at the moment, and a hiring freeze is never a hiring freeze, we do a lot of recruiting, mostly replacement recruiting. But there's no budget right now to implement anything. And furthermore, I think we need to fix the basics first. Fix all the basics that are part of the recruiting process before we start implementing any AI.

[00:06:38] Adriaan: Fair enough. Have you seen for yourself, because you do use Claude, any specific use cases where AI makes you more productive as a head of TA, especially in an organization that isn't immediately adopting AI itself?

[00:06:58] Karim: Well, it helps me in selling or building the storyline that I need to sell the business case of a centralized TA function. It helps me tell that story or build that storyline. That's one use case I've used it for recently. We're not using it in recruitment, so we're not using it to screen any resumes or anything like that. But we do use it, for example, in building interview structures for certain roles. It's very helpful for the recruiters, but also for the hiring managers, to have a structured or semi-structured flow to your interview process. And that works well as long as you know how you need to structure it. You shouldn't rely on the prompt that is immediately answered.

[00:07:46] Karim: Maybe first I've got a question for you, Adriaan. I have AI fatigue. I reckon I'm not the only one here. What's your take on that?

[00:07:55] Adriaan: So I am on the other side of AI fatigue. I love it. Every day I'm spending more and more time with AI. But what I've noticed is that all of a sudden we have long email chains from Claude, ChatGPT or whatever, which on the surface feel like a really good argument or thought process. Typically people spend some time prompting it, going back and forth until the answer is to their liking, and then they send these pretty long emails to our management team, for example. What I noticed is that the minute I see this is written by ChatGPT, I either stop reading it or I take it less seriously, even though I've done the same myself. I've spent time crafting an email and then run it through ChatGPT and asked it to enrich it, structure it in a better way. But because it feels AI, it feels like someone has not really done the work. So that's where I really recognize the AI fatigue. The second thing is, for a time I wrote some articles that I first wrote myself and then used AI to polish up. I did 95% of the work, and 5% was done by AI to weed out my grammar mistakes. But it actually became less readable because of it. You might have seen a post or an article from me where it smells like AI. The minute that happens, you lose your audience. That's something I really relate to. On the flip side, we are doubling down on AI. We are looking at every process: what can we automate, where can we make the life of our sourcers, recruiters, and coordinators better? And for me overall, we're a hundred-plus people company right now, and there's a lot of things that I'm not aware of. So how can I get the data from the people that are working for our clients presented in a digestible way with predictive values, so that I see: on this account we are on track, on this account we're not, and here are maybe some of the reasons why. That to me is interesting to see, how fast that goes and what we're capable of doing.

[00:10:19] Karim: Yeah, I think that makes sense. I've had those situations as well, where you're basically communicating with somebody that's only using AI, and you take them less seriously, or at least whatever topic you're talking about, you don't want to spend time on. But there's so much efficiency and it adds so much value at the same time. It's a balance.

[00:10:40] Adriaan: Yeah. Our head of marketing, Andrew, who you also know: last year we spent 4,000 or 5,000 euros to build our TRC website, our recruitment conference website. We built it from the ground up. Designer, developer, a really nice looking website. This year we built the entire website ourselves with Lovable. I dare to say that it might even look better than it does now. It took Andrew, if I'm not mistaken, three to four days to build the website, and you can argue whether that is the best use of his time, but he didn't use a designer, didn't use an engineer, and the entire website is running. And now what's interesting is that we are revamping our entire website, and that estimate was 15,000 to 20,000 euros, because there's a lot of content on our website. The podcast is there, a lot of videos, a lot of blogs, so the backend is actually a lot more work. It's not just a WordPress anymore. And Andrew decided: I'm going to redesign the website completely myself, based on his experience designing the TRC website, because now with AI he's capable of doing it. This is a bigger project. He's two weeks in and he's hands down doing everything himself. But the beauty is, because he understands now how to design the entire website that's going live next week, we are able to create specialized landing pages for prospects, specialized landing pages for specific communities that we're developing. We're able to tailor the content to the specific target audiences that we want to attract. And besides the hours of Andrew, we're not spending any money on it.

[00:12:24] Karim: Wow. So it's not only cheaper, it also adds so much more to the end product that you want to have.

[00:12:32] Adriaan: 100%. And also the speed. It just took him two weeks. Of course, this is two weeks of him not doing other stuff. But our current website we designed probably four years ago, and that was a project of four or five months before we finally went live.

[00:12:47] Karim: Yeah, that's pretty cool.

[00:12:49] Adriaan: Hey, we're going to talk about the Recruitment Blueprint. Tell me a little bit more: what is this project, how do you feel about it, and give us some more context.

[00:13:01] Karim: The way that I've seen recruitment being done in some companies is that they have a hiring challenge, and the first thing that they do is set up the team. They hire the recruiters, they get the LinkedIn seats, they basically get an ATS, and then they're like, "Okay, let's hire." But what I've found in doing that, or what I've seen in other companies, is that after a while you see that every manager hires differently. Nobody agrees on ownership. There's still a ton of agencies everywhere within the company being used. Processes are inconsistent. So I feel like you've missed a few steps if you start with "let's get recruiters on board." And I think that's what the blueprint is about: taking ten steps back and thinking about the direction that you want to head towards, instead of "okay, let's build a recruitment team."

[00:13:57] Adriaan: Let's make this a little bit more concrete, because you're working on your current blueprint within BDR Thermea. Okay, blueprint, that sounds obvious. What does that look like, why is it important, and how can it be useful?

[00:14:12] Karim: Well, to give you some context, we're in a transformation. We have a certain ambition and goal for the next four years, with a lot of strategic...

[00:14:21] Adriaan: Give our listeners a little bit more context. How big is BDR Thermea?

[00:14:27] Karim: We are about 7,000 people around Europe. Our focus is Europe only. There are about seven countries where the majority of those people sit, but also smaller hubs in Poland and so on. We grew through acquisition, meaning that we were quite decentralized and still are. Every country has its own management team, managing director, and so on. A couple of years ago a new CEO came in and set up a new transformation, a new strategy that we call Focus, Transform, Perform. That strategy is cascaded into the company, and right now we're focusing on the target operating model: how should every department be organized in order to achieve that strategy? Recruitment is part of the bigger HR target operating model, and that's the exercise we're doing right now. That's the context. The reason why I don't want to start with "okay, let's build the team first" is that within that bigger transformation and strategy, I want to have a recruitment puzzle piece that explains what we actually want recruitment to be. What do we want talent acquisition to be, and how is it going to solve the issue that is relevant for the strategy? I feel like a lot of people underestimate that question, because not every recruiter is equal. Not every recruitment team is equal. A recruiter can be a scheduler, it can be an advisor, or it can be somebody that does strategic workforce planning. That vision part explains who owns what, what does good hiring look like, what is talent in the context of our organization, everything around talent acquisition governance, but also what hiring managers can expect from talent acquisition and vice versa. So that is the bigger direction we're thinking about, and we're starting with governance first. Very boring processes: who does what, who owns what, this is how it should be. That consists of process descriptions, but also hiring guides. And the hiring guides are where the nuance lies, in how a recruiter has the freedom to do the job according to what they believe is best. Those are a few puzzle pieces that we're trying to fit into the bigger puzzle.

[00:16:50] Adriaan: Because you currently have recruiters in all these different countries. Is the idea behind your strategy that you're going to centralize the recruiters in one central hub?

[00:17:01] Karim: Not per se. I think centralization in itself is not a goal, and there's a lot of local nuance. Fully centralizing is not ideal, but neither is doing everything locally the same way. The question is: what generally benefits from consistency? What can we do in the same way, what should be country-specific and what shouldn't? When it comes to everything I just mentioned, the governance, the reporting, at least the hiring standard, that should be the same. Where you have the differences could be the labor markets, the way of communicating, how manager relationships should be done locally. So the centralization part is not a goal in itself. It's more about alignment, let's call it intelligent alignment. That's the right way I see it.

[00:17:56] Adriaan: How do you start a project like this? Because it sounds logical on paper, but the reality is usually much messier.

[00:18:04] Karim: There's already momentum where every department needs to look at their target operating model, how they should be structured. That's already a mandate, so that's momentum we can use where the decision has already been made. We want the Ulrich model: global HR, different centers of excellence or expertise, and then local HR business partners.

[00:18:30] Adriaan: Give a little bit more context about that model.

[00:18:32] Karim: It's an old-fashioned model. It's basically a centralization of HR, where you have HR leadership and then your centers of expertise or centers of excellence: compensation and benefits, recruitment, people operations, where everything is done according to a similar consistency and standard, instead of having everything decentralized. That's something we're heading into. The complexity is that we're in a lot of different countries with different sizes of HR and different maturity of HR. The first thing we did is the direction: this is what recruitment should be. The second part is actually building the business case and setting a baseline. What does recruitment look like in each of the countries? How many hires do we do? How is hiring being done at the moment? So that we have one slide deck of how hiring currently looks in all of the countries. Using that baseline, we can slowly work towards what it should look like. In practice that might be some global recruiters and local recruiters, where I also believe in a flexible recruitment part: I would say 60 to 70% we can do ourselves, and the other 30 to 40% is RPOs, agencies, things like that.

[00:20:00] Adriaan: Do you work closely with the head of HR on this program, or do you still need to do a lot of convincing internally to get this alignment?

[00:20:10] Karim: Yeah, definitely. The whole centralization and Ulrich model is something that was talked about before I joined two years ago. So it is about getting people on board with it, and getting local HR on board with it. Right now we say we want to do this, but the moment that you actually change, that's when the real resistance obviously comes up.

[00:20:31] Adriaan: That makes a lot of sense. You've worked at software companies, you've worked at physical product companies. You mentioned earlier that industry doesn't really matter, but if you look at your current role and the roles you've come from, what are some of the misconceptions around context and companies?

[00:20:57] Karim: I think every company thinks they're uniquely complicated. That kind of sums it up for me. Every company thinks they're unique, every company thinks they're the best, or whatever. But I've worked at companies like VanMoof, a recruitment software company, and now at a bigger company. It has exactly the same hiring problems. Every hiring manager cliche that you can think of exists at BDR. On one hand it's frustrating to deal with it again, but on the other hand you can have a lot of impact. Whether I recruited software engineers at VanMoof, or tripled the size of Recruitee's team, or now in a manufacturing environment, it feels exactly the same.

[00:21:49] Adriaan: But it's interesting, because Recruitee was a couple hundred people when you left, and BDR Thermea is 7,000 people, very international, different types of roles. What has been one of the transitions you had to make in your way of working, and maybe thinking, from a fast-growing tech scale-up to a more established corporate company?

[00:22:12] Karim: I think things move slower. And when I say things move slower, I mean that the thinking is as fast, it's just the decision-making that is a little bit slower. So I have to really pace myself. At Recruitee, when I was Global Head of TA there, when I wanted to implement a certain process or tool, I could think about it on Monday morning, and either in the afternoon or on Tuesday I'd start the implementation phase. Here you need to think very critically about what you want to do, build the business case, get the people on board, talk about it, have a lot more meetings, and then maybe you get the mandate to do it or not. That can take weeks or months.

[00:23:00] Adriaan: Yeah, I think it's inevitable. That's interesting. For you, in terms of the blueprint, that's something you're actively working on. Do you have longer timelines for this, or is this something that is going to be executed this year? And does technology and AI, sorry to bring it up again, play a role?

[00:23:24] Karim: What we want to do in the summer period is work on the baseline, what I mentioned. Then by the end of the year we want to have at least the governance in place: how we hire and what this looks like. And from 2027 onwards: okay, if we're going to centralize it, what would it look like in terms of the people, and what's the specific team we would need? When it comes to the AI piece, that's a different story. There won't be any budget to implement any AI, but what we can do is have principles in place, so that when we're ready to start thinking about AI tooling, we know how we want to use it internally. What we should automate and what we should not automate. There are a lot of AI tools that take over the work from the recruiter, and that might work in, perhaps, a retail setting, but in a more senior and executive hiring setting that might be different. So we're going to look at the principles and decide how we're going to work with AI, and then pause that until we're ready to actually implement something.

[00:24:33] Adriaan: Fascinating. Even if you can make the argument of "hey, we're going to be more efficient, we can do more with less", is that a case to be made, or not really?

[00:24:44] Karim: It is up for discussion, but as I mentioned earlier: let's fix the basics first. There are so many hiring managers in the company that it's going to take some time to ensure we have a process defined, but also having it actually working in practice as we want it to work. That's going to take some effort. We also use a lot of agencies, so we need to consolidate our agency use on each level, from executive to more specialized. There's already so much we can do right now to ensure that everything is in place, and then we can make it more efficient and better with AI.

[00:25:26] Adriaan: So good. Karim, where can people follow you and follow your journey?

[00:25:32] Karim: I need to post more on LinkedIn, I confess. You might see one or two posts of me where you think, hey, that might be written by AI. But I'm going to post more on LinkedIn about all my challenges and learnings in recruitment and other things. So follow me on LinkedIn.

[00:25:50] Adriaan: We'll put your name in the show notes. Karim, thank you so much for being a guest. I love to follow your journey. Let's definitely have a catch-up next year to see where we are in this journey, in terms of localized, centralized, what is happening with technology and AI, and where you are on that journey.

[00:26:10] Karim: Sounds good. Thanks so much.

[00:26:11] Adriaan: Thanks so much, Karim.

[00:26:12] Karim: Cheers.