Swiss Re’s Andreas Berger Sees AI as the Next Step in Understanding Risk
-
Swiss Re’s Andreas Berger Sees AI as the Next Step in Understanding Risk
Swiss Re’s Andreas Berger Sees AI as the Next Step in Understanding Risk -
AWS’s Matt Wood on Why AI Assumptions Need an Expiration Date
AWS’s Matt Wood on Why AI Assumptions Need an Expiration Date -
Heidelberg Materials’ Dominik von Achten on AI’s Ultimate Advantage
Heidelberg Materials’ Dominik von Achten on AI’s Ultimate Advantage -
Reckitt’s Kris Licht Believes AI Transformation Starts with People
Reckitt’s Kris Licht Believes AI Transformation Starts with People
Swiss Re’s Andreas Berger Sees AI as the Next Step in Understanding Risk
A conversation with BCG CEO Christoph Schweizer and Swiss Re Group CEO Andreas Berger.
Swiss Re, one of the world’s largest reinsurers, is bringing AI into the core of its 162-year-old business. In underwriting and claims, the company aims to speed up routine work so its people can spend more time with clients and make better decisions about risk.
Group CEO Andreas Berger recently sat down with BCG CEO Christoph Schweizer to make the case that growth is AI’s most undervalued benefit. The following interview has been edited for concision.
Christoph Schweizer: You’ve always stood out to me as a very growth-oriented leader. Now you’re CEO of Swiss Re, and here comes AI. What does AI mean for Swiss Re, and how do you reimagine your core business?
Andreas Berger: Our purpose is to make the world more resilient. That means we need to understand risk and risk exposures. We collect data, we model, and we produce risk insights. AI is just the continuation of that work. It’s not an experiment for us.
Reimagination is the key word here. We are reimagining our core processes, using agentic AI in particular. That will have an impact on our operating model, to the benefit of our customers, because we will get quicker and closer to them.
Swiss Re has invested in data foundations systematically for a long time. How do you think about data architecture and governance, and all that comes with that?
Our industry is known for fragmented IT landscapes and a lot of IT legacy, which, to be frank, stands in the way of being AI-ready. We identified this problem eight or nine years ago and started changing the way we organize data and building a globally integrated tech infrastructure.
Our foundation is an end-to-end ontology and a clear data strategy, complemented by that infrastructure. The integration matters because each and every AI use case is built into our data strategy and tech, so the benefits come through. That is what I call “AI readiness” from a company perspective.
Let’s talk about people. How do you ensure that they are on the journey and excited about it?
We only have people. We have data. Yes, you could say that we have technology and a balance sheet. That’s it. So people are the most important aspect—and we always focus on the human in the loop. We don’t take away responsibility from the people. People make decisions. AI is not making decisions for people. I think that’s a very crucial element in our approach to AI.
We opened AI to the whole Swiss Re population. Every employee can use ChatGPT and Copilot. But to use these tools, they need to get ready, so we train and prepare them so that AI becomes second nature. We also spend a lot of time on our core job families. For us, it’s underwriting and claims, as well as operations, finance, and control functions like risk management. People who work in those job families need to imagine how AI will change their world.
Take construction engineering underwriting. For a big project, going from submission to quote can take up to three months. With agentic AI, we could reduce the 25 subprocesses to 5 and cut the time to quote to one day. What are the benefits? It’s not reducing cost. It’s freeing up time for underwriters to spend with the customers, assess and price risk, and find ways to prevent it. That’s significant.
One way to use that productivity gain is to reinvest it in growth. How do you think about AI and new opportunities for growth at Swiss Re?
I think growth is the most underestimated and undervalued case. In our reinsurance business unit, we do single-risk underwriting, which we call facultative reinsurance. That unit receives roughly 100,000 submissions a year, and with all the unstructured data involved, handling that volume is almost humanly impossible. Each submission looks different, whether it arrives as an email or a note from a phone call, so a lot of time goes into making each one ready to underwrite.
Agentic AI can eliminate those constraints. If an underwriter could previously handle, say, 100 submissions, the time freed up means that the underwriter could now handle 400. That’s real growth. On top of that, underwriters can spend more time creating new underwriting solutions for clients.
One thing Swiss Re has always had is strategic clarity about what makes it the better partner for customers. How do you look at strategic clarity in the world of AI?
When it comes to benefits, there are two buckets we look at. One is productivity. The second is better decision-making. We can see patterns in data that help our claims people with fraud detection, for instance, or with recoveries through what we call subrogation.
These aspects are beneficial for us internally, but they’re also beneficial for the clients. Through data platforms, we make our risk insights and models available, so customers can increase their awareness. With more granular data, they can make their own decisions about risk financing and build resilience.
The reality of AI is that there are a lot of unknowns, from how the technology will evolve to what it means for regulation. What will the company look and feel like as you deal with them?
The unknowns are our business. If things were predictable, you wouldn’t need reinsurance. For our customers, we see new, emerging risks with AI, such as malfunctioning software and algorithms, and also false positives. These new exposures will come our way, and we need to make sure we understand them.
You can see it with data centers. Demand for insurance and reinsurance capacity is going through the roof. The problem is the concentration of risk in a single location, with asset values that are immense—in the double-digit billions.
When you bring all this together, it’s an important reimagination of what Swiss Re does and how your people work. What does that imply for your own leadership philosophy?
It’s a tough one at times, because you face a dilemma. You see the beauty and the excitement in AI. At the same time, people fear that their job might be gone or become so different they can’t imagine it. As a leader, I think you must focus on three things.
Number one, you have to anticipate. What will underwriting look like in the future? You have to detach yourself from current processes to imagine it. That’s not easy. Second, you need self-motivation as a leader, because people will always tell you it’s not possible, or that we’ve done it this way for many years, so why change?
Third, and probably most important, is inspiration. You need inspiration to take people along on the journey, because people will be skeptical. Look for digestible, manageable milestones where you can see the successes coming through. We don’t always have all the answers, so at times you have to feel your way into it. But the conviction needs to be there.
We call this video series “AI Pioneers,” and Swiss Re and you are AI pioneers. What have you learned that would be useful for other executives?
First, I learned you must be fast—and have a level of conviction to be fast—because a lot of people will tell you it’s not possible or that the investments are too high. You need a leap of faith at times. Fail fast, and when the outcome isn’t what you expected, correct course. If it still isn’t working, stop it. That kind of discipline is very important.
When I set out an ambition for benefits such as growth, our people are skeptical because they haven’t seen it yet, so their own estimates of the benefits are cautious. But in every use case I’ve seen so far, the impact has been bigger than people estimated. That’s good news, and it encourages people to embrace AI.
Swiss Re, one of the world’s largest reinsurers, is bringing AI into the core of its 162-year-old business. In underwriting and claims, the company aims to speed up routine work so its people can spend more time with clients and make better decisions about risk.
Group CEO Andreas Berger recently sat down with BCG CEO Christoph Schweizer to make the case that growth is AI’s most undervalued benefit. The following interview has been edited for concision.
Christoph Schweizer: You’ve always stood out to me as a very growth-oriented leader. Now you’re CEO of Swiss Re, and here comes AI. What does AI mean for Swiss Re, and how do you reimagine your core business?
Andreas Berger: Our purpose is to make the world more resilient. That means we need to understand risk and risk exposures. We collect data, we model, and we produce risk insights. AI is just the continuation of that work. It’s not an experiment for us.
Reimagination is the key word here. We are reimagining our core processes, using agentic AI in particular. That will have an impact on our operating model, to the benefit of our customers, because we will get quicker and closer to them.
Swiss Re has invested in data foundations systematically for a long time. How do you think about data architecture and governance, and all that comes with that?
Our industry is known for fragmented IT landscapes and a lot of IT legacy, which, to be frank, stands in the way of being AI-ready. We identified this problem eight or nine years ago and started changing the way we organize data and building a globally integrated tech infrastructure.
Our foundation is an end-to-end ontology and a clear data strategy, complemented by that infrastructure. The integration matters because each and every AI use case is built into our data strategy and tech, so the benefits come through. That is what I call “AI readiness” from a company perspective.
Let’s talk about people. How do you ensure that they are on the journey and excited about it?
We only have people. We have data. Yes, you could say that we have technology and a balance sheet. That’s it. So people are the most important aspect—and we always focus on the human in the loop. We don’t take away responsibility from the people. People make decisions. AI is not making decisions for people. I think that’s a very crucial element in our approach to AI.
We opened AI to the whole Swiss Re population. Every employee can use ChatGPT and Copilot. But to use these tools, they need to get ready, so we train and prepare them so that AI becomes second nature. We also spend a lot of time on our core job families. For us, it’s underwriting and claims, as well as operations, finance, and control functions like risk management. People who work in those job families need to imagine how AI will change their world.
Take construction engineering underwriting. For a big project, going from submission to quote can take up to three months. With agentic AI, we could reduce the 25 subprocesses to 5 and cut the time to quote to one day. What are the benefits? It’s not reducing cost. It’s freeing up time for underwriters to spend with the customers, assess and price risk, and find ways to prevent it. That’s significant.
One way to use that productivity gain is to reinvest it in growth. How do you think about AI and new opportunities for growth at Swiss Re?
I think growth is the most underestimated and undervalued case. In our reinsurance business unit, we do single-risk underwriting, which we call facultative reinsurance. That unit receives roughly 100,000 submissions a year, and with all the unstructured data involved, handling that volume is almost humanly impossible. Each submission looks different, whether it arrives as an email or a note from a phone call, so a lot of time goes into making each one ready to underwrite.
Agentic AI can eliminate those constraints. If an underwriter could previously handle, say, 100 submissions, the time freed up means that the underwriter could now handle 400. That’s real growth. On top of that, underwriters can spend more time creating new underwriting solutions for clients.
One thing Swiss Re has always had is strategic clarity about what makes it the better partner for customers. How do you look at strategic clarity in the world of AI?
When it comes to benefits, there are two buckets we look at. One is productivity. The second is better decision-making. We can see patterns in data that help our claims people with fraud detection, for instance, or with recoveries through what we call subrogation.
These aspects are beneficial for us internally, but they’re also beneficial for the clients. Through data platforms, we make our risk insights and models available, so customers can increase their awareness. With more granular data, they can make their own decisions about risk financing and build resilience.
The reality of AI is that there are a lot of unknowns, from how the technology will evolve to what it means for regulation. What will the company look and feel like as you deal with them?
The unknowns are our business. If things were predictable, you wouldn’t need reinsurance. For our customers, we see new, emerging risks with AI, such as malfunctioning software and algorithms, and also false positives. These new exposures will come our way, and we need to make sure we understand them.
You can see it with data centers. Demand for insurance and reinsurance capacity is going through the roof. The problem is the concentration of risk in a single location, with asset values that are immense—in the double-digit billions.
When you bring all this together, it’s an important reimagination of what Swiss Re does and how your people work. What does that imply for your own leadership philosophy?
It’s a tough one at times, because you face a dilemma. You see the beauty and the excitement in AI. At the same time, people fear that their job might be gone or become so different they can’t imagine it. As a leader, I think you must focus on three things.
Number one, you have to anticipate. What will underwriting look like in the future? You have to detach yourself from current processes to imagine it. That’s not easy. Second, you need self-motivation as a leader, because people will always tell you it’s not possible, or that we’ve done it this way for many years, so why change?
Third, and probably most important, is inspiration. You need inspiration to take people along on the journey, because people will be skeptical. Look for digestible, manageable milestones where you can see the successes coming through. We don’t always have all the answers, so at times you have to feel your way into it. But the conviction needs to be there.
We call this video series “AI Pioneers,” and Swiss Re and you are AI pioneers. What have you learned that would be useful for other executives?
First, I learned you must be fast—and have a level of conviction to be fast—because a lot of people will tell you it’s not possible or that the investments are too high. You need a leap of faith at times. Fail fast, and when the outcome isn’t what you expected, correct course. If it still isn’t working, stop it. That kind of discipline is very important.
When I set out an ambition for benefits such as growth, our people are skeptical because they haven’t seen it yet, so their own estimates of the benefits are cautious. But in every use case I’ve seen so far, the impact has been bigger than people estimated. That’s good news, and it encourages people to embrace AI.
AWS’s Matt Wood on Why AI Assumptions Need an Expiration Date
A conversation with BCG Managing Director and Senior Partner Dylan Bolden and AWS Chief AI and Technology Officer Matt Wood.
Amazon Web Services (AWS), a leading cloud provider, has made computing power available to any business that needs it. The company now aims to do the same for AI and is investing heavily to meet demand that already exceeds capacity.
Chief AI and Technology Officer Matt Wood is helping lead that push. He recently sat down with BCG’s Dylan Bolden to talk about why AI’s next, faster phase will reward leaders who build small, multiskilled teams and give them the freedom to set their own course. The following interview has been edited for concision.
Dylan Bolden: Matt, how do you move forward when there’s so much uncertainty around what the next model development is going to be? How do you talk to your customers about that?
Matt Wood: If you take a step back, the past couple of years have felt very fast moving—and they have been. New models, approaches, and agents arrive virtually every single day. I’m sure that by the time you and I finish chatting, there’ll be something else out there worthy of our attention. Who is ahead changes often and dramatically.
But if you look at progress in capability, it has been metronomic and linear, up until about three months ago. At that point, we started to see early indicators that we’re entering a completely different phase of AI acceleration. We’re likely to see self-optimizing and self-improving AI systems where the AI system itself is also the scientist that builds the next version of itself.
With that acceleration, the next few years will be turbulent. Organizations will need a willingness to hit the reset button, deciding at a rapid clip whether to keep a project going or stop it.
Part of that discipline is putting an expiry date on your assumptions. If you think a particular approach is the right one, or that using AI is too expensive for another use case, you’re probably right. But you’ll probably be wrong in 6 to 12 months. Build mechanisms into your organization to reassess those assumptions. You may net out at the same place, but the reassessment has to happen.
As a leader, how are you thinking about leading within AWS and helping your customers see through the hype and deal with the pressure to deliver results?
There are a lot of first-order decisions that leaders can help teams make. You can help teams form in ways that are much more diverse than in the past. You want engineering close to the business, and you want legal close to the engineering and the business. Product management has a slightly different role to play going forward. Building those multifaceted, multiskilled teams is something leaders can help with. Those teams will also tend to be much smaller, which is great because you can run more of them in parallel. Managing that is a leadership challenge and a skill we all need to develop.
Those teams operate very differently. They do a lot of upfront work before you see any delivery. With AI, slow is smooth, and smooth is fast. The teams can spend a lot of time designing, specifying, and going deep on the details where previously they may have spent their time implementing. Successful teams tend to fall in love with the problem instead of the solution. If you’re focused on the problem, anything that helps you solve it is beneficial, even if that means throwing work away, resetting, or following a squiggly line to success.
Being responsive as a leader and providing air cover for those teams is a critical component of leadership. Inside large organizations, the instinct is often to reach for more command and control and more inspection. Sometimes the opposite is more beneficial. Give the teams more agency, more freedom, and broader remits and let them control their own destiny. Teams with that control are much more impactful.
The obvious question is, what’s next? What’s next for AWS? What’s next for AI?
So far, with AI, we have some smash-hit use cases. There are maybe a dozen, including coding assistants and personal assistants, cybersecurity assistants, protein folding, and gene regulation. Very exciting, very impactful. But the truth is that for almost all AI usage today, demand exceeds the capacity available to build the capabilities people have in mind.
So we’ll invest about $200 billion across Amazon this year. We’re building more servers. We have doubled the amount of compute capacity on AWS since 2022, and we’ll double it again in the next year.
Building a new AI system is going to be as cheap, as easy, and as frequent as building a website. That future may still be ten years away. It could be ten months away. Think back 30 years to the web. There were probably dozens, but not hundreds, of impactful, useful websites and a smaller number of killer applications, such as search and e-commerce. Today there are millions of websites and apps covering a remarkably broad set of capabilities. Part of our job at AWS is to bring that future state as close to tomorrow as possible.
Amazon Web Services (AWS), a leading cloud provider, has made computing power available to any business that needs it. The company now aims to do the same for AI and is investing heavily to meet demand that already exceeds capacity.
Chief AI and Technology Officer Matt Wood is helping lead that push. He recently sat down with BCG’s Dylan Bolden to talk about why AI’s next, faster phase will reward leaders who build small, multiskilled teams and give them the freedom to set their own course. The following interview has been edited for concision.
Dylan Bolden: Matt, how do you move forward when there’s so much uncertainty around what the next model development is going to be? How do you talk to your customers about that?
Matt Wood: If you take a step back, the past couple of years have felt very fast moving—and they have been. New models, approaches, and agents arrive virtually every single day. I’m sure that by the time you and I finish chatting, there’ll be something else out there worthy of our attention. Who is ahead changes often and dramatically.
But if you look at progress in capability, it has been metronomic and linear, up until about three months ago. At that point, we started to see early indicators that we’re entering a completely different phase of AI acceleration. We’re likely to see self-optimizing and self-improving AI systems where the AI system itself is also the scientist that builds the next version of itself.
With that acceleration, the next few years will be turbulent. Organizations will need a willingness to hit the reset button, deciding at a rapid clip whether to keep a project going or stop it.
Part of that discipline is putting an expiry date on your assumptions. If you think a particular approach is the right one, or that using AI is too expensive for another use case, you’re probably right. But you’ll probably be wrong in 6 to 12 months. Build mechanisms into your organization to reassess those assumptions. You may net out at the same place, but the reassessment has to happen.
As a leader, how are you thinking about leading within AWS and helping your customers see through the hype and deal with the pressure to deliver results?
There are a lot of first-order decisions that leaders can help teams make. You can help teams form in ways that are much more diverse than in the past. You want engineering close to the business, and you want legal close to the engineering and the business. Product management has a slightly different role to play going forward. Building those multifaceted, multiskilled teams is something leaders can help with. Those teams will also tend to be much smaller, which is great because you can run more of them in parallel. Managing that is a leadership challenge and a skill we all need to develop.
Those teams operate very differently. They do a lot of upfront work before you see any delivery. With AI, slow is smooth, and smooth is fast. The teams can spend a lot of time designing, specifying, and going deep on the details where previously they may have spent their time implementing. Successful teams tend to fall in love with the problem instead of the solution. If you’re focused on the problem, anything that helps you solve it is beneficial, even if that means throwing work away, resetting, or following a squiggly line to success.
Being responsive as a leader and providing air cover for those teams is a critical component of leadership. Inside large organizations, the instinct is often to reach for more command and control and more inspection. Sometimes the opposite is more beneficial. Give the teams more agency, more freedom, and broader remits and let them control their own destiny. Teams with that control are much more impactful.
The obvious question is, what’s next? What’s next for AWS? What’s next for AI?
So far, with AI, we have some smash-hit use cases. There are maybe a dozen, including coding assistants and personal assistants, cybersecurity assistants, protein folding, and gene regulation. Very exciting, very impactful. But the truth is that for almost all AI usage today, demand exceeds the capacity available to build the capabilities people have in mind.
So we’ll invest about $200 billion across Amazon this year. We’re building more servers. We have doubled the amount of compute capacity on AWS since 2022, and we’ll double it again in the next year.
Building a new AI system is going to be as cheap, as easy, and as frequent as building a website. That future may still be ten years away. It could be ten months away. Think back 30 years to the web. There were probably dozens, but not hundreds, of impactful, useful websites and a smaller number of killer applications, such as search and e-commerce. Today there are millions of websites and apps covering a remarkably broad set of capabilities. Part of our job at AWS is to bring that future state as close to tomorrow as possible.
Heidelberg Materials’ Dominik von Achten on AI’s Ultimate Advantage
A conversation with BCG CEO Christoph Schweizer and Heidelberg Materials CEO Dominik von Achten.
Heidelberg Materials, a global giant in heavy building materials and solutions, recently embarked on a bold effort to future-proof its 152-year-old business. A key part of this journey is embedding AI into its core physical operations, including plans to roll out more than 100 autonomous vehicles by 2028.
CEO Dominik von Achten is leading this charge, taking a highly scalable approach to how AI can drive top-line growth and secure a competitive edge. The following interview has been edited for concision.
Christoph Schweizer: Dominik, technology and sustainability are central to Heidelberg Materials’ impressive results. You have a very particular approach to growth. Tell us more about that.
Dominik von Achten: I think for Heidelberg Materials, our approach is first profitability, and then it’s transformation along two lines: sustainability and automation through AI. I think we have to secure the first to get to the second. The industry had some challenges in the past, in terms of being a big carbon emitter and being one of the most offline businesses on the planet. And where there are challenges there are big opportunities, and at Heidelberg, we grab them.
You say you are one of the most offline businesses out there, yet you decided to take on AI with an enterprise-wide approach. What drove that decision, and how is that working now?
We traditionally very much thought in pilots, but we realized that if you really want to change things you need an approach that’s scalable. There are big pockets along our entire value chain primed to be addressed with automation through AI. We can apply it everywhere—starting in the quarries, moving through the labs and production sites, and extending to logistics, sales and distribution, and the customer interface. The upside of being so offline is that we have countless opportunities to implement AI. We pilot, but with an immediate plan for scale. For example, in North America, we now run 12 plants remotely from a single location. I think we are the first company piloting at this scale in North America, and our goal is to apply this model to most of our operations globally.
One of our convictions at BCG is that enterprise AI transformation requires the CEO to lead by example. What are you doing to show your team that AI is a priority and that you’re personally involved?
Well, the good news is I don’t need to signal much. I’m really excited myself. To put that enthusiasm into action, we created formats like the CEO AI hour, where I connect directly with colleagues around the world who have built something with AI, not just pitched an idea. It cuts through the traditional hierarchy. While it shows my personal commitment, the real goal is to prove that in a company of 50,000 people, AI empowers everyone.
After a record 2025, Heidelberg Materials is leading the industry. How does AI factor into your broader financial and strategic philosophy?
Initially there was this idea that AI could be used purely for efficiency gains, but I think we probably underestimate the speed and precision it can bring. For us, the real differentiator is top-line growth. I’m pushing hard to use AI on the commercial side and in product development because it can completely transform the customer experience.
As for the workforce, I don’t buy the narrative that it will just eliminate jobs. Instead, AI agents will work side by side with our people to elevate what they can do. Ultimately, the biggest unanswered question isn’t about job loss but ROI: will the massive gains in speed and efficiency outweigh the new costs of running the AI itself?
Heidelberg Materials embraces several flavors of AI: predictive, generative, agentic, and now physical. How do you think about these different arenas, and where do things go from here?
Physical AI really comes to life in our ready-mix plants. Concrete mix designs are essentially like recipes, and we have thousands of them per plant. What we’ve done to connect this physical world with AI is collect 180,000 historical recipes from around the world to build a unified platform.
We are now using this massive data set to train algorithms that optimize the quality, safety, and margins of our mixes. Ultimately, the future is about taking the data the physical world gives you and using AI to leverage it into better products.
You are combining new AI technology with your historical knowledge of these materials. Is that where your real competitive advantage lies?
Exactly. It all comes down to what we call domain know-how. For example, in our R&D centers and our labs, we typically have 30-day, 60-day, or 90-day testing periods to see if the concrete curing is right. But with all the historical data we’ve assembled, we can now train algorithms to simulate that outcome with extreme precision and speed. I am convinced it will eventually free us from needing to do those physical tests altogether. That is a huge improvement, and we can only do it because we have that historical domain data.
To close out, let’s bring this back to the bottom line. You’re focusing this technology on the core of your day-to-day operations rather than on just support processes. How does that ultimately impact your P&L?
There are certainly opportunities in administrative processes, and I don’t want to downplay that. But that’s something everybody can do. By applying AI directly to the core processes of our value chain, scale and differentiation become the two levers that will secure Heidelberg Materials’ competitive edge.
Heidelberg Materials, a global giant in heavy building materials and solutions, recently embarked on a bold effort to future-proof its 152-year-old business. A key part of this journey is embedding AI into its core physical operations, including plans to roll out more than 100 autonomous vehicles by 2028.
CEO Dominik von Achten is leading this charge, taking a highly scalable approach to how AI can drive top-line growth and secure a competitive edge. The following interview has been edited for concision.
Christoph Schweizer: Dominik, technology and sustainability are central to Heidelberg Materials’ impressive results. You have a very particular approach to growth. Tell us more about that.
Dominik von Achten: I think for Heidelberg Materials, our approach is first profitability, and then it’s transformation along two lines: sustainability and automation through AI. I think we have to secure the first to get to the second. The industry had some challenges in the past, in terms of being a big carbon emitter and being one of the most offline businesses on the planet. And where there are challenges there are big opportunities, and at Heidelberg, we grab them.
You say you are one of the most offline businesses out there, yet you decided to take on AI with an enterprise-wide approach. What drove that decision, and how is that working now?
We traditionally very much thought in pilots, but we realized that if you really want to change things you need an approach that’s scalable. There are big pockets along our entire value chain primed to be addressed with automation through AI. We can apply it everywhere—starting in the quarries, moving through the labs and production sites, and extending to logistics, sales and distribution, and the customer interface. The upside of being so offline is that we have countless opportunities to implement AI. We pilot, but with an immediate plan for scale. For example, in North America, we now run 12 plants remotely from a single location. I think we are the first company piloting at this scale in North America, and our goal is to apply this model to most of our operations globally.
One of our convictions at BCG is that enterprise AI transformation requires the CEO to lead by example. What are you doing to show your team that AI is a priority and that you’re personally involved?
Well, the good news is I don’t need to signal much. I’m really excited myself. To put that enthusiasm into action, we created formats like the CEO AI hour, where I connect directly with colleagues around the world who have built something with AI, not just pitched an idea. It cuts through the traditional hierarchy. While it shows my personal commitment, the real goal is to prove that in a company of 50,000 people, AI empowers everyone.
After a record 2025, Heidelberg Materials is leading the industry. How does AI factor into your broader financial and strategic philosophy?
Initially there was this idea that AI could be used purely for efficiency gains, but I think we probably underestimate the speed and precision it can bring. For us, the real differentiator is top-line growth. I’m pushing hard to use AI on the commercial side and in product development because it can completely transform the customer experience.
As for the workforce, I don’t buy the narrative that it will just eliminate jobs. Instead, AI agents will work side by side with our people to elevate what they can do. Ultimately, the biggest unanswered question isn’t about job loss but ROI: will the massive gains in speed and efficiency outweigh the new costs of running the AI itself?
Heidelberg Materials embraces several flavors of AI: predictive, generative, agentic, and now physical. How do you think about these different arenas, and where do things go from here?
Physical AI really comes to life in our ready-mix plants. Concrete mix designs are essentially like recipes, and we have thousands of them per plant. What we’ve done to connect this physical world with AI is collect 180,000 historical recipes from around the world to build a unified platform.
We are now using this massive data set to train algorithms that optimize the quality, safety, and margins of our mixes. Ultimately, the future is about taking the data the physical world gives you and using AI to leverage it into better products.
You are combining new AI technology with your historical knowledge of these materials. Is that where your real competitive advantage lies?
Exactly. It all comes down to what we call domain know-how. For example, in our R&D centers and our labs, we typically have 30-day, 60-day, or 90-day testing periods to see if the concrete curing is right. But with all the historical data we’ve assembled, we can now train algorithms to simulate that outcome with extreme precision and speed. I am convinced it will eventually free us from needing to do those physical tests altogether. That is a huge improvement, and we can only do it because we have that historical domain data.
To close out, let’s bring this back to the bottom line. You’re focusing this technology on the core of your day-to-day operations rather than on just support processes. How does that ultimately impact your P&L?
There are certainly opportunities in administrative processes, and I don’t want to downplay that. But that’s something everybody can do. By applying AI directly to the core processes of our value chain, scale and differentiation become the two levers that will secure Heidelberg Materials’ competitive edge.
Reckitt’s Kris Licht Believes AI Transformation Starts with People
A conversation with BCG CEO Christoph Schweizer and Reckitt CEO Kris Licht.
Reckitt, a global manufacturer of everyday health and hygiene products, recently embarked on a bold effort to simplify its business by focusing on core brands. A key part of this journey is embedding AI into its “Fuel for Growth” strategy.
CEO Kris Licht is spearheading this shift. He recently sat down with BCG CEO Christoph Schweizer to talk about taking a deliberate approach to how AI can create value. The following interview has been edited for concision.
Christoph Schweizer: Kris, you made a decision that AI would be central to your organization’s growth strategy. When was that point, and what triggered it?
Kris Licht: When ChatGPT was released in late 2022, I was running our global health business and was not yet CEO. I took my leadership team to San Francisco, and we spent a week immersing ourselves, figuring out what the technology could do and where it might go. That was a pivotal moment. We decided this was one of the few things worth going fast on—worth real investment and real time. And it was important that we engaged everyone in the journey.
How did you decide the best way to frame the program, what function to focus on, how much experimentation to allow, and how rapidly to scale?
The first choice was to go fast. And the second decision—and I think this is the most important choice we made so far—was not to go out and buy whatever people were selling us.
We took a first principles approach instead, and said, “We want this to change the way our whole organization works.” To figure out what that would look like, we used good old time and motion studies and mapped the work people were doing in a given domain. Before we deployed a tool, we studied what people did, where effort was being spent, or where handoffs or complexity slowed things down.
That has turned out to be a profound exercise. It helps us understand where AI can create value by enabling a different way of working.
You made the choice to go right into the heart of what makes you win, starting with marketing and R&D versus a support function. What drove that choice?
We thought these tools would change the game on what “good” looks like and what excellence requires. Marketing drives our growth, and R&D drives our innovation, so the stakes were highest there. It was counterintuitive, but starting in those domains forced a more serious conversation. It was not simply, “Where can we use AI?” but “How are we trying to change the company?” Once we had successes in these harder domains, it proved something to the organization: if it works in marketing and R&D, it can work anywhere.
One of the things I’ve observed in my own company and at many of our clients is that people say, “Yeah, we tolerate failure, and there will be some of that.” But reality can feel different. How did you manage this at Reckitt and evolve the culture?
You have to keep people focused on the long term. One use case may fail. One piece of the investment may not generate a big return. But as long as, in aggregate, you’re moving forward, you’re on the right path.
I’ve found empathy to be one of the big unlocks. You give the model a little space, you interact with it more productively, and you are less disappointed when it doesn’t do exactly what you wanted the first time. Ownership also matters. The executives who deployed AI in their functions own it because they cocreated the solution. The foundational work they’ve done is now helping us move faster in later waves of expansion through the company.
AI models get better with every release. So the question isn’t about the technology itself, it’s about how your organization embraces it. That has big implications for talent. How are you approaching this and ensuring your teams build the right expertise?
One of the things that we have to figure out is, “How do we create mastery in our more junior people?” In many industries, including ours, junior people previously developed mastery by doing a lot of the basic work.
But AI is reducing the most repetitive and least interesting parts of work, which makes many jobs better and the output more consistent. But this means we have to think harder about career paths and capability building. How do people learn more about AI and its capabilities while simultaneously developing deep functional expertise? We have not fully cracked that yet, but it is one of the most important questions.
Empathy, judgment, and an ability to bring people along are differentiating factors in an organization’s success. You’ve been at the forefront of this AI transformation. What advice would you give to other CEOs who are trying to move beyond experimentation?
Be clear about the impact you want to create, then work backward. What work has to change? What analysis is needed? What design choices have to be made? What needs to happen for the value to show up?
There are many ways to approach AI, and I would not claim there is only one right path. But for us, the first step is a hard conversation about how you want to change your company. Plenty of people have strong products, but buying technology comes far down the line.
What has helped you lead this as CEO?
Curiosity! I found myself in a panel presentation about how AI enabled the discovery of entirely new compounds. It immediately captured my imagination and made me wonder what that kind of capability could do for Reckitt.
When things get hard or when we have to revisit our assumptions, I continually go back to that moment. It reminds me that this is worth doing despite all the challenges.
Reckitt, a global manufacturer of everyday health and hygiene products, recently embarked on a bold effort to simplify its business by focusing on core brands. A key part of this journey is embedding AI into its “Fuel for Growth” strategy.
CEO Kris Licht is spearheading this shift. He recently sat down with BCG CEO Christoph Schweizer to talk about taking a deliberate approach to how AI can create value. The following interview has been edited for concision.
Christoph Schweizer: Kris, you made a decision that AI would be central to your organization’s growth strategy. When was that point, and what triggered it?
Kris Licht: When ChatGPT was released in late 2022, I was running our global health business and was not yet CEO. I took my leadership team to San Francisco, and we spent a week immersing ourselves, figuring out what the technology could do and where it might go. That was a pivotal moment. We decided this was one of the few things worth going fast on—worth real investment and real time. And it was important that we engaged everyone in the journey.
How did you decide the best way to frame the program, what function to focus on, how much experimentation to allow, and how rapidly to scale?
The first choice was to go fast. And the second decision—and I think this is the most important choice we made so far—was not to go out and buy whatever people were selling us.
We took a first principles approach instead, and said, “We want this to change the way our whole organization works.” To figure out what that would look like, we used good old time and motion studies and mapped the work people were doing in a given domain. Before we deployed a tool, we studied what people did, where effort was being spent, or where handoffs or complexity slowed things down.
That has turned out to be a profound exercise. It helps us understand where AI can create value by enabling a different way of working.
You made the choice to go right into the heart of what makes you win, starting with marketing and R&D versus a support function. What drove that choice?
We thought these tools would change the game on what “good” looks like and what excellence requires. Marketing drives our growth, and R&D drives our innovation, so the stakes were highest there. It was counterintuitive, but starting in those domains forced a more serious conversation. It was not simply, “Where can we use AI?” but “How are we trying to change the company?” Once we had successes in these harder domains, it proved something to the organization: if it works in marketing and R&D, it can work anywhere.
One of the things I’ve observed in my own company and at many of our clients is that people say, “Yeah, we tolerate failure, and there will be some of that.” But reality can feel different. How did you manage this at Reckitt and evolve the culture?
You have to keep people focused on the long term. One use case may fail. One piece of the investment may not generate a big return. But as long as, in aggregate, you’re moving forward, you’re on the right path.
I’ve found empathy to be one of the big unlocks. You give the model a little space, you interact with it more productively, and you are less disappointed when it doesn’t do exactly what you wanted the first time. Ownership also matters. The executives who deployed AI in their functions own it because they cocreated the solution. The foundational work they’ve done is now helping us move faster in later waves of expansion through the company.
AI models get better with every release. So the question isn’t about the technology itself, it’s about how your organization embraces it. That has big implications for talent. How are you approaching this and ensuring your teams build the right expertise?
One of the things that we have to figure out is, “How do we create mastery in our more junior people?” In many industries, including ours, junior people previously developed mastery by doing a lot of the basic work.
But AI is reducing the most repetitive and least interesting parts of work, which makes many jobs better and the output more consistent. But this means we have to think harder about career paths and capability building. How do people learn more about AI and its capabilities while simultaneously developing deep functional expertise? We have not fully cracked that yet, but it is one of the most important questions.
Empathy, judgment, and an ability to bring people along are differentiating factors in an organization’s success. You’ve been at the forefront of this AI transformation. What advice would you give to other CEOs who are trying to move beyond experimentation?
Be clear about the impact you want to create, then work backward. What work has to change? What analysis is needed? What design choices have to be made? What needs to happen for the value to show up?
There are many ways to approach AI, and I would not claim there is only one right path. But for us, the first step is a hard conversation about how you want to change your company. Plenty of people have strong products, but buying technology comes far down the line.
What has helped you lead this as CEO?
Curiosity! I found myself in a panel presentation about how AI enabled the discovery of entirely new compounds. It immediately captured my imagination and made me wonder what that kind of capability could do for Reckitt.
When things get hard or when we have to revisit our assumptions, I continually go back to that moment. It reminds me that this is worth doing despite all the challenges.
