The AI Question HR Keeps Getting Wrong: Syed Ali Abbas on Impact Over Efficiency at HelloFresh

Show notes

In this episode of the Leaders in Talent podcast, host Adriaan Kolff interviews Syed Ali Abbas, Global Vice President of People for Business Units at HelloFresh Group, the parent company of HelloFresh and Factor. The company employs almost 20,000 people worldwide, with a people team of over 400. Before HelloFresh he was Executive Director of HR at AT&T, CHRO at PacNet Group, Group HR Director at Global Fashion Group, and Chief Strategy Officer at Epitome Global. He also invests in early-stage HR startups and performs stand-up comedy.

Most companies ask how AI can make HR more efficient. Abbas asks a different question: how does AI change the business, therefore the organisation, therefore HR? He explains what that reframe means in practice for HRBPs, talent acquisition and L&D, why cost-led AI business cases stall after year one, and why some companies that cut customer service teams hired them back six months later. He also walks through the HelloFresh production centre where AI, robotics and supply chain software were combined so that packing productivity doubled without replacing anyone, and one pilot centre now offers 350 meal options instead of 50. Plus his four-hour Claude workforce planning experiment, and why he is not worried about his own AI imposter syndrome.

Timecodes

01:36 Welcome and Abbas's background

02:35 A year of stand-up, and roasting HR from the inside

06:29 The question most companies ask about AI, and the better one

07:18 FOMO, token maxing, and "what's the plan and what's the payoff?"

09:31 What the reframe means for HRBPs, talent acquisition, and L&D

13:08 Inside HelloFresh: anonymous AI self-assessment and partner workshops

15:02 Tool access, the IT and legal review team, training the leaders first

17:36 HelloFresh in context: 20,000 people, a 400-person people team

18:14 Incremental mindset vs exponential mindset

20:37 Why the companies that cut deepest hired people back six months later

22:14 "Technically I'm pretty basic": Abbas on his own AI journey

23:56 The four-hour Claude workforce planning experiment

26:34 Peak hype, the trough of disillusionment, and not panicking

32:19 The installed base problem, and why people are the hardest part

34:21 Pull factors and push factors in a transformation

35:44 AI, robotics and 350 meals: inside the HelloFresh production centre

40:32 How to reach Abbas

___________________________

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 Syed Ali Abbas: https://www.linkedin.com/in/syedaliabbashr/

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

___________________________

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

Transcription

[00:01:36] Adriaan: All right, ladies and gentlemen, welcome to another episode of the Leaders in Talent podcast. I'm happy today to be joined by Syed Ali Abbas. Abbas, welcome to the podcast. Great to have you. Abbas is the Global Vice President of People for Business Units at HelloFresh Group, which is the parent company of HelloFresh but also Factor. Abbas previously held other leadership positions, for example Executive Director of HR for AT&T, CHRO for PacNet Group, Group HR Director for Global Fashion Group, and Chief Strategy Officer for Epitome Global. Next to leading global HR teams, Abbas also invests in early-stage HR startups, and Abbas also does stand-up comedy. So I think we're going to have fun today, Abbas.

[00:02:30] Syed: We are, we are. I've been looking forward to this one, Adriaan. Thank you for having me.

[00:02:35] Adriaan: So good. And I think we talked about it, right? I did my first stand-up comedy a couple of weeks ago, and it's not easy. It was much harder than I anticipated to really tell a story in a way that your audience is not only engaged, but actually laughs.

[00:02:54] Syed: For sure. It can be a lonely place when it's not going well, but it's super fulfilling when it goes well. I actually finished my first year in stand-up a couple of weeks ago. There was a particular organization which has a people leaders day. It's a VC fund. They invite 80 or 90 of their VPs of talent and chief people officers every year for a day of learning. And last year, I closed that show with a stand-up set. It was my first. I got so much encouragement, so much love, and so much positivity from that, that it gave me the courage to actually pursue this as a passion. And it was really nice that at the end of my first year, I went back into the same event and shared new jokes with the same people, and it was great fun once again.

[00:03:41] Adriaan: So good. What are some of the stereotypes around HR, or around people?

[00:03:46] Syed: Are you sure you want to do this? Because what I do is I roast HR people quite a bit.

[00:03:55] Adriaan: So good. It's important, right? Let's not take ourselves too seriously.

[00:04:00] Syed: Yeah. Never let it be said that I'm not balanced. So I usually always start by roasting the TA folks. They're always the early adopters. An example I can give you is how this whole TA space has been evolving so much in the past few years. It was talent acquisition, then it became talent. I was a keen student of this space, and I was looking at it and I was like, "Oh, so they've changed the name to talent. What are they doing differently?" Not that much. And then it became talent mobility, and it was just talent, but internal. Then it became talent success, and there were just a few onboarding slides added on. Then it became AI talent, and, oh, that's just using Metaview for note-taking. And then it became skills talent, skills-based talent, skills-driven talent. I was like, "If you weren't interviewing for skills till now, what the hell were you interviewing for?" And that's the kind of gentle but insider joke I usually make. To be fair, I have to talk about the HRBPs if you'll give me a minute, so there's balance in the world. With the HRBPs, we always talk about them being so mysterious and impressive, walking from conference room to conference room, having strategy meetings with the business leaders. But as somebody who's on the inside, I can tell you the reality is one business leader will say, "Hey, my expense voucher is stuck with finance. Can you please go and check why it's not clearing?" And the other one will say, "I hired this amazing software engineer from the US. They're moving to Berlin. Can you please make sure their dog has a comfortable journey and clears the immigration?" And the third person says, "You know the high performer I told you about last week? They're now a low performer. Can you please fire them, and figure it out in European labor law?" And that's the kind of stuff that we talk about. Funnily enough, when I was doing this joke at a show last year about the immigration for the dog, two HRBPs came up to me. They were from a big fintech company, which I shall not name here. They actually came to me and they said, "Hey, we actually had to do that."

[00:05:59] Adriaan: So good. And so relatable. I love it. It's important that we have these conversations. There's a big gap to fill, so I think you're going to get booked much more often. If anything, for the people listening to this podcast, reach out to Abbas, not only for great HR advice, but also if you want to have a laugh and want to be entertained, and also have someone hold a little mirror to our profession, right?

[00:06:27] Syed: Thank you.

[00:06:29] Adriaan: That being said, Abbas, before we were recording this call, we had a little bit of a brainstorm: "Hey, what can we talk about?" Because there are so many avenues. I was almost reluctant to go down the AI path, because so many people are talking about AI. But inevitably, when we were talking, I realized, okay, you actually have something interesting to say, especially how you look at AI from an HR leadership standpoint and some of the things that are happening within the organization. One thing that stood out is that many organizations ask how AI can make HR more efficient. But during our conversation, you ask a different question: how does AI change the business and therefore the organization, and then what does that mean for HR? Can you explain to me the difference in how you think about that?

[00:07:18] Syed: Yeah, absolutely. This statement, when I make it internally in the company last year and now externally this year, always gets me a few strange looks, because everyone is in this FOMO mode of "Hey, let's take Claude or ChatGPT or whatever and create workflows and notebooks and chatbots and agents and all this stuff," and everyone's busy doing it. My view on this is sure, we can do that, but what's the plan and what's the payoff? And this is not just a people function question, it's a question for every part of the business. Because right now we hear about all those horror stories of token maxing and people flying through their budgets, using up all their budget four months into the year and not really having a lot of ROI. Everyone's struggling to say, "Okay, what did we get out of this expense that we have?" For me, there are two key questions to ask, whether it's the people team or any other team who's doing this, and I'll get to the people team in a second. The key question in general to ask is: how can we leverage AI to have the biggest impact on the business? And impact doesn't just mean visible impact. It also means outcomes. It means return on investment for the money that you put in, because otherwise you can put it somewhere else and do something good for the customers or the employees. This is the same question for every technology. So what are these areas? In general, the biggest impacts, the biggest needle movers, are always things that either relate to the customer, or relate to the product that the customer consumes, or relate to the value chain, which is how you create that product and then deliver it to the customer. And of course, in terms of outcomes, especially for for-profit businesses, it's about the top line and the bottom line, the revenue and the profit. So for me, that is the key question. And when you do that, there's a natural thing that happens, which is that you then get focus. So you're not going out and doing a thousand experiments. You're actually looking at a few areas, focusing on them and doubling down.

Now, what does this mean for the people team? It means different things for different parts of the people team. So if I'm the HRBP team, and I'm leading that, creating a chatbot or creating a notebook, yeah, okay, it'll make a few things convenient. It'll make me do something a bit faster, make me share a report faster with my business stakeholder. But the reality of it is, if I'm also focused, just like the business leaders, on customer outcomes, on product improvement, on supply chain improvement, I'm actually going to sit with the business leader and say, "Okay, if you're going to drive all this, and you're going to change all this with AI, then how can I help you using my skills as an HRBP to reinvent the organization, reinvent the structure of who's doing what, where in this chain? What is the culture we need to make this happen and to pivot to a new way of doing things? What are the ways of working at the very granular level that we have to change? And how do we make this happen for the business?" That's the HRBP piece. But if I'm sitting in talent acquisition or in people ops, a lot of my roles are not really looking at working with the business stakeholders on that kind of stuff. They're really focused on employees and taking them through the talent life cycle. So for me then, if I'm in talent acquisition or people ops or even EX, I can say, "Hey, maybe I need to use AI to make my talent processes better or faster or both, so I do a better job of attracting, motivating, retaining the talent that is coming into the company to achieve all these things that the business is working on. And maybe I do it in a way which frees up more time of people to do that work rather than to get caught up in bureaucracy forms, self-service in traditional legacy ERP systems, which people get lost in, and so on." And for L&D, the answer will be completely different, because for L&D I would say let's sit with the people who are the thought leaders on what we're going to do with AI and with data in the company, what our direction is, our vision is as a company, and then let's make sure we upskill people to the level each of them needs to actually do this. So the level of upskilling needed for a software engineer who is a front-end engineer but now is trying to migrate to becoming a full stack engineer, or a tech product person whose role is changing dramatically and who has to start making MVPs instead of writing their product requirements document, that requirement is very different compared to what I do for an accounts payable person or a payroll person. So I need to find that upskilling and bring it there so that the rest of the business is enabled, again, in the service of those big, huge focus areas for the business. That aligns us, that makes us relevant for the business, that gets us the seat at the table, and that makes us a part of the AI journey rather than just creating another chatbot for onboarding.

[00:13:02] Adriaan: Do you have an example where this concretely materialized?

[00:13:08] Syed: Yeah, sure. What I'm sharing with you as a model is pretty much how we are looking at it in HelloFresh. So our L&D team, for example, has not tried to upskill everyone. They have set a lowest common multiple bar and said, "You know what? Everyone needs a certain level of AI in their roles." We give them a sort of self-assessment which they can use. It is not tracked. It's completely anonymous. They can download it on their laptop and use it, so nobody knows, including their manager, where they are on AI, so it's a safe space. And then based on that, there are recommended learning journeys that are available to upskill yourself on the basics, AI 101. So it's the kind of stuff made publicly available by OpenAI, Anthropic, AWS, Google, et cetera. But then what we do is specifically for the areas where we need to upskill people in certain tracks, we bring that to play. So if somebody is working on the Amazon tech stack, the AWS tech stack, we will bring people from Amazon to have conversations with them. If somebody is using the analytics suite that we have and is now trying to figure out how to do this using the Google platform, we will actually bring people from Google in. And then the Claude guys are coming in and saying, "We want to focus on taking the experience and the vision of our business leaders and helping them quickly trial ideas, using information from across the company." So the Claude folks will come in and do a workshop around how to use their main LLM, on how people in the top management can quickly synthesize information and use it for scenarios, for decision-making, et cetera. So I could go on for hours because we have thought through this quite clearly, but that's very much a live example.

[00:15:02] Adriaan: How do you do that internally with your own team? Have you given everyone access to Claude or OpenAI? What is the mandate that you've given people? Where are you on that journey?

[00:15:13] Syed: So there are a couple of parts to it. One is that self-discovery part. The second part of it is we actually suggested to all the people in the company last year that if you see a really cool AI tool out there, and the big ones we already have via Gemini and Claude, but if you see something really cool there that's a point solution or that is specific to your space, you can actually submit it to a core team where there's an IT person and a legal person sitting there to evaluate it and to make sure that if we sign a contract with them, it protects our customer and employee information, and it is done in a safe space. And our work does not become their product in the future, because that's the nature of LLMs. They train on your information, and then they create something with that. We also made sure that the leaders themselves are trained quickly so they can role model this. And on top of all of that, within the people team and specifically my team, we were looking at how do we progress people beyond the generic training. So I enrolled some of the folks in AI boot camps with certain organizations which are more for HRBPs. But we tested them. We didn't go all in. We tried it with a few people first to see if it was good enough quality, because everyone is an AI trainer these days, but a lot of the stuff out there is really poor quality. And then we also did some work ourselves. But most importantly, we told them, "Hey, use it, get familiar with it, get it into your day-to-day work, and after a couple of months it'll become second nature." Then all this talk about everyone has to build workflows, and every org chart will have agents on it, this is all very nice theoretical conversation, which makes everyone get excited and drives up adoption. That serves the vendors who are selling this to us. But for us, what matters is we use it for our work, and we start using it today. And for that, you just need to develop the habit.

[00:17:08] Adriaan: Hey, you mentioned don't waste time building AI workflows, but sit with the business and drive culture change. Tell me a little bit more about that, and how do you instruct your teams to do that with the business?

[00:17:22] Syed: The culture change. HelloFresh is what people call a digital native company, which basically means we are very much a tech-first company, even though we are selling a...

[00:17:36] Adriaan: Abbas, in terms of size, give our listeners a little bit: how big is HelloFresh, and how big is your HR organization?

[00:17:43] Syed: Right. We are almost 20,000 people around the world, primarily in developed markets rather than emerging markets. The people team is over 400 people. More than half of that is actually sitting in our operation centers, which is where the food comes in and gets packed into boxes or gets cooked and then gets shipped out, and then the logistics for that. And then the rest of it is HQ folks who are doing the business partner, rewards, learning, et cetera, like myself.

[00:18:14] Syed: In terms of the culture change, I think it's got a few angles to it. One is the mindset. In terms of mindset, there is an incremental mindset, which is what has traditionally been the case in enterprise tech. When you roll out a big ERP system, you say, "Okay, there's a bunch of transactions. Now I'm going to automate those transactions. I'm going to force other people to do self-service. Theoretically, it's going to save me 10 headcount around the world who are doing whatever transaction in people, finance, whatever, analytics, and that is a saving of 800,000 euros per year, and that goes straight to my bottom line after we deduct the license fees of the tool and the cost of maintaining it." I won't go into the details of why that doesn't really work well and why most of those fail. There's a reason for that, which people who have bought these tools before know. But if you take that traditional mindset, a lot of companies which you see stumble a little bit in their first experiments with AI have actually taken this mindset, which is the guys who say, "I scrap my entire people team. I scrap my entire customer service team." It's all agents and robots and androids or whatever name they want to come up with. There are two problems with that. Number one, AI is still in its early stages. The big models have already ingested everything that is on the internet. They're now creating copies of that with changes so that they can keep creating more material for it to train on, while we all onboard onto it. But the reality of it is that these things are still totally dependent on people to provide context beyond the basic script and beyond the basic A, B, C, D in your prompt. So they're not really that autonomous right now. They're still work in progress. But the bigger problem is that when you go with a purely cost-based business case like this, you make your CFO and your CEO happy in the first year because you have saved X number of euros. It goes straight to the bottom line. But you can't keep firing people at that rate every year, and all the money and all the time that you have sunk into that is something that is not used for driving the big picture stuff that I was talking about before: supporting the customer outcomes, supporting the value chain outcomes, all that stuff. So for me, the companies which have not done well are exactly those who fire all the customer service people and then, oh wow, six months later they have to hire back a bunch of them because it didn't work, but also in the long run it wouldn't have worked anyway.

The companies that do well are the ones which have more of an exponential mindset, and they say, "I'm not going to use..." This is still an enterprise tech for us. It's not that different from an SAP or a Salesforce or a Workday. But the unique characteristics of this allow me to do some things differently, and that will drive exponential gains for me, and that means I can drive revenue outcomes and profitability outcomes. I can make customers happy and drive efficiency and productivity. I can launch new products and make more money out of the existing products and grow them more. And this stuff, generally those big topics, they are usually cross-functional, and that is why a lot of the existing enterprise tech, which is very functional in nature, they have a finance module, people module, whatever, that's why it doesn't play so well. AI is a very good example of a tech which actually can bring together the data across all these things and drive improvements across all of them. So for me, the companies that do better are going to be those companies which can really think with that exponential mindset and go cross-functional across entire value chains, and that's where the difference comes of not just making your function more efficient or your people team more effective, but the entire company grows.

[00:22:14] Adriaan: How do you do that for yourself? What has been your own personal journey and how do you look at that? Have you immediately gone into trying these tools and just prompting and seeing where this goes? Are you part of networks where you get inspired? Where do you get your knowledge from, and how do you think and train yourself to stay up to date?

[00:22:32] Syed: So if you look at technical proficiency, I'm pretty basic. I can do a lot of stuff with prompts, but I'm not at the sort of place where I'm building agentic workflows and agents, 50 agents are working for me and all that stuff, but that's because I chose not to invest time on it. I think nine years ago, I was using UiPath for robotic process automation and using it to ingest data from 10,000 resumes in 30 seconds. And I've been in the tech space from an HR perspective for 20-plus years, so it's not that I can't do it, but I chose to spend my time and invest it more where I thought it would have more value, which is, in my role, I'm the person who should be laying out the vision and laying out the strategic options for the business and for my team to say, "Here's how we can use AI. Now let's think about actual use cases for HelloFresh in the real world where we're working, and if we need a little bit of help to get the tech solution over the line, let's just go to someone in our tech team or let's go to one of the software providers and actually get their help." So for me, I would say technically I'm pretty basic, but in terms of the actual ability to think through topics and solve problems, I'll give you an example.

I was at the same event I was talking about, where I did the stand-up. The earlier part of the day, I had my work hat on, so I was doing a lab on workforce planning and people analytics for a group of CPOs and VPs of talent who were attending. And this was basically an open discussion around how does this change in the age of AI, et cetera. So the day before that I said, "Look, it's a bit silly for me to go into that and just talk theory. Let me sit down and just use Claude and see what I can do." So I set myself a target of four hours. In those four hours, half an hour of that was spent with me watching a circle in process because Claude wasn't connected to my Google Sheets. So I only realized after multiple tries that I needed to download it to Excel and then upload to Google Sheets. So yes, as I said, very basic on the tech side. But in the rest of the time, I was able to create dummy employee data for a fictional B2B software company which exists in Europe, operates in multiple locations, has X revenue, X number of employees, X amount of cash burn, and X amount of profitability. And then I was able to seed certain assumptions into it, ranging from performance to AI adoption levels to other stuff. And then on top of that, I introduced some constructs there, which were basically dummy salary bands for those people, and a premium for AI people because they're hired at a higher salary, and some outcomes in terms of employee engagement if the founder of that company went too aggressive with AI and suddenly chopped a bunch of people and did that again the next year. And that was my scenario one. And my scenario two was there was a much more thoughtful and focused approach in terms of workforce changes. Some people left, but some joined, and they built a sort of self-reinforcing machine. And then I was able to run simulations on those scenarios, and Claude was using Python on the back end, so I didn't need to use Python. And then I was able to take those simulations and forecast attrition, hiring, workforce cost, revenue growth, profitability, cash burn, and even if the company would need to raise new funding two years later in scenario one versus being able to invest and be self-sufficient in scenario two. All of this in three and a half hours.

[00:26:29] Adriaan: Yeah, no, the power of AI is pretty remarkable.

[00:26:33] Syed: Right?

[00:26:34] Adriaan: Because that's something that I'm part of, like different entrepreneur groups, different founders, and everyone is playing around in some sort of capacity with AI. And that's where I get a lot of my inspiration from, and how other people have looked at their business, how other people have automated certain steps. Because there's so much to do. Is there something or somewhere that you go where you get your information, or is it just thinking business logic 101? It's like, hey, in terms of the business, what does the business need? To your point, where does this business want to go? And how does HR then play a transformational role in there, and are there factors in that HR field where AI can be of benefit?

[00:27:23] Syed: Look, I'm a huge believer, and I think this is the benefit of having worked in companies that are really small, like 20, 25 people, Epitome was that, and like AT&T, which is 250,000 people, and having seen waves of tech change over the years. So honestly, I think I've been very lucky to have that kind of diversity of experience and just sheer number of years of experience, which is quite nice now, although I still won't tell you my age. What has really helped me is not to panic and to take a very systematic step-by-step approach, because we're in the middle of what's called the Gartner hype cycle. Right now, AI is at peak hype, and we can already see the early signs of naysayers now starting in the past few weeks because of various things that have happened. We don't need to go into them. But essentially, now it's going to go into what's called the trough of disillusionment, when everyone realizes, "Oh God, it's just another enterprise tech play where they were trying to get us into the ecosystem so we can't get out, because our data is trapped there, and now they can increase the license fees." And then it will level out to a business as usual mode in a couple of years. Because of having seen this happen before, I've not panicked. I've not gone out and started doing prompting courses and day and night doing AI experimentation on everything I'm doing. I've actually been able to take a step back and say, "Look, what does the business actually need in those heavy-hitting topics, and how can we use AI to help? And how can we use it in the people function?" And then take a very sensible return on investment driven approach.

And honestly, the most difficult part of it is not me feeling that I have imposter syndrome of some sort on AI. I'm okay. My worry is everybody else in the world seems to have it, and they're all in a rush to do stuff. And even in HelloFresh, where we've had a very sensible, step-by-step approach to this, there are still people who are like, "We're moving too slow. We need to go faster. We need to do this and that." And I'm like, "But how do we have... Forget about the money. How do we have the mental bandwidth and the actual work capacity from the right people to achieve all these things?" We have to prioritize. We have to pick what are the winners long term for us and then double down on it. So yes, I speak to people who are much smarter than me on this topic. I follow some people on LinkedIn. Luckily, I'm able to have a good quality selection in my network, so I know who's actually talking sense and who's just boosting AI for whatever reason, or a particular tool. And beyond that, it's also the experience from the past and seeing if I can learn from history, not just from what I've done and the companies I've been in, but in general, learn from history about what has worked well and what hasn't worked well in these kinds of transformational tech waves, like cloud computing, all the way back to the internet in our lifetimes. It's been great fun. I'm really enjoying it. And it's very fulfilling because it gets you to do new things, and you do them and you're like, "Oh, okay, I'm not ready for retirement. I'm not even hitting mid-career. I'm actually able to keep up with all the young people who are AI native, and I can add value to them just like they add value to me."

[00:30:48] Adriaan: Is that something that you have to remind other leaders of as well, that are like, "Let's go. This is the future"? There's definitely those, and I probably count myself into that camp as "Let's give everyone access to AI right now. Let's go." I'm very red in my personality, so it's like this is the future, right?

[00:31:07] Syed: You're a founder. It's your job to do that. It's your job to have the vision and push everyone towards that vision. So it's totally understandable. We're the same. I think there are leaders in HelloFresh who have that vision, and they push, and it's their job. And others who are there saying, "Hey, wait a minute, that's fine. Let's do it. But can we pick three out of five, or five out of 10, so that we have a better chance of success?" I'm fairly lucky to be surrounded by people here who are quite sensible about AI, so we have not run out of our AI budget. We are not telling our people to token max, or that everyone has to create an agent. So I think it's a good environment that I'm in. I think the biggest challenge I have is actually more of a leadership challenge, which is that when you have these waves of transformation, and usually it's a number of things, it's not just AI. Our biggest win in AI, we can talk about it if we have time left, in terms of how we are doing things, is actually not AI by itself. It's integrating AI into everything else in the company, and then we have that exponential gain. So I can talk about that if you like, but...

[00:32:18] Adriaan: Yeah, yeah.

[00:32:19] Syed: Before we get to that, I think the biggest challenge is that you have an installed base in the company. You have software, you have processes, you have capital investment, you have physical goods, you have offices, you have people. Now, the thing with software is you can change it. You can recontract. Process, you can re-engineer the process, you can remap it. CapEx, you can get out of something and invest in something else. You just have to be smart around it. Your CFO will guide you through it. The problem comes when it comes to the people side of it, because the people who are successful in your company today, pre-AI or pre any transformation, are successful because they're successful with the way things work today. They're not an exact fit for how things are going to work tomorrow. So now when you start pushing to change things and see how they're going to work tomorrow, I think there's a massive friction that happens where some of them will be able to pivot, especially leaders, because they have the most installed base of experience, and good habits and skills and personalities. So some will succeed, some will fail, some will drop out because they don't want to try. This part is very difficult, and for me it's always the same. Even though AI is a different type of transformation, in every transformation it's always the same thing. This is the most difficult part of the job, for me personally, and in general I think for people people, because you can't just magically wave a Harry Potter wand and say, "Presto, you're now an AI person." People are people. They take time to learn. They take time to pivot. Some of the habits are inbuilt over twenty years. It's stuff they've done since they were in school. They're not going to change some of those habits.

So you just do your best to make that period of friction be less painful rather than more painful. So you create a safe space. You actually give people enablement to catch up. You give people the opportunity to develop themselves before you start performance managing them on something new so big. And then those are the pull factors, where you're pulling them to you and to the change. Then you have to do the push factors, and you say, "Look, good habits, product person, no more product requirements document. Show me an MVP." That's your habit change. Software engineer, you've got to do this. HR person and BP, you've got to do this. That's the push part, where you push them towards building those habits and pivoting. And ideally, your pull and push is of a balance that works for as many people as possible, to bring them along with you on that transformation journey. And that is tough. I never underestimate that. I know people team folks like me, we all fixate on change management as being the most important thing. It's a great skill. It's like project management. And change management is not the be-all and end-all of everything. But when you dig deeper within that change management, this part, that friction of the overlap, that part is the one that can make or break your transformation, especially at leadership level.

[00:35:39] Adriaan: And you said you had some kind of example on how you're doing it at HelloFresh?

[00:35:44] Syed: So I think one of the things that people talk about is AI is very much this, you take AI and you replace a person, and that's your game. The classic enterprise tech model. So one thing that we're doing very interestingly is, in our markets, we generally have been in the past shipping between 30 to 50 recipes onto the app for customers to look at. Then they choose and they order a couple for themselves, their family, whoever. And then it's delivered to them within a week from the time they order, and that's the subscription model. It works. And we have many, many happy customers, millions of happy customers with that around the world. Now, to reinvent this, if we had taken that approach of saying, "Hey, the people who are packing the boxes in the production center, we replace them with robots to pack the boxes," that might make it a bit faster, 10% faster. And then we take the people who are taking the pictures of the recipes and then writing the recipe in HelloFresh language, and then printing the cards for our customers and uploading it on the app. If we automate all of that with AI, it'll make it faster. Now, all of these things will give you incremental gains. And there is a high chance of failure because AI may have glitches. It may not be that mature in some areas.

Now, what we have done is we took some of those ideas of automation. So the recipe writing, we've written recipes for 15 years, and we have 15,000 recipes in our data. I don't know how many thousand anyway, in our database. AI now has enough to do a good job writing it in multiple languages. The photo studio touching up, yeah, AI can do it quite quickly. We don't need 10 people doing it. We need a few people doing it. But here's the interesting part. That does not move the needle. You can say, "Okay, I don't backfill a couple of people who resign. I save a couple of headcount, a couple of hundred thousand euros." Where you really move the needle is at the customer level, in terms of customer delight or speed or outcomes or cost. So what we have done is we actually, over time, took AI and have been cleaning up our data, because ultimately AI is a data play as much as it is a tech play, and used that to better forecast what customers like in different parts of the world and what their ordering patterns are, and really understand in every geography that we operate in, not countries, but cities, what is the best case scenario for customers, what they like, what they don't like, including their personal preferences. We stack rank them. But then also we invested in robotics, we invested in supply chain software, and then of course we invested in AI, and we pulled that together.

The outcome of that is that now all our markets, instead of thirty to fifty recipes, are able to present between fifty to a hundred recipes to customers already, with an incremental gain. However, our future state is currently being piloted. There are thirty thousand customers on it in a particular production center in the US where we are trialing this. In that center, instead of the employees walking around and picking up the ingredients and packing them, they actually have the conveyor belt bringing some ingredients to them, robots bringing other ingredients to them, mobile robots, and the shelves behind them are stacked based on intelligence of which are the most likely ingredients to be used based on the most likely recipes to be ordered. And with that, the number of people needed to pack a box falls in half because it takes half the time to pack the box. And the person is not tired. They don't have accidents running around, tripping, falling, trying to get stuff in a hurry. So our people are happier. They're not being replaced by robots. Their productivity has doubled. And based on that, that particular production center is able to offer those thirty thousand pilot customers three hundred and fifty meals as options, and those are now stack ranked by AI based on personal preferences, geographic preferences. And our goal is to get to five hundred. So essentially, it's like having twenty restaurants in your pocket now, where you can order a wide variety of food for all sorts of dietary choices, family choices, et cetera. And that's a game changer. And we would not do that if we had not taken all those different pieces and then put AI in there as one of the pieces of the puzzle, rather than just using it to replace people.

[00:40:32] Adriaan: Wow. What a great example. That's very concrete. I love it. Abbas, so helpful, so great to speak. What is the best way for people to get in touch with you if they want to book you for stand-up comedy or want to talk about it or just follow your journey, about what you're doing at HelloFresh? Is it LinkedIn, or is there any other website?

[00:40:51] Syed: Yeah, I'm on LinkedIn every day. Look, I think because of the nature of this conversation, I'm sharing a lot of things, but I'm very happy when people reach out to me and share what they're doing, because we all learn together and rise together. And especially I think I'm seeing some amazing things happen on the tech side of HR. Some of the things I'm seeing in talent acquisition now, mind-boggling. I couldn't even have dreamed of those things one or two years ago. So please, to all your listeners, if you have something cool to share, please share. I would love to learn and have a chat over a coffee, virtual or otherwise.

[00:41:30] Adriaan: Amazing. Abbas, thank you so much for joining the podcast. This was super interesting.

[00:41:35] Syed: Thank you, Adriaan. Take care. Talk soon.