From Chemical Engineer to Chief Digital Officer - How Evonik Builds an AI-Ready Culture Across 33,000 Employees
Guest: Henrik Hahn, Chief Digital Officer at Evonik
Hosts: Seth Earley, CEO at Earley Information Science
Chris Featherstone, Sr. Director of AI/Data Product/Program Management at Salesforce
Published on: August 29, 2022
In this episode, Seth Earley and Chris Featherstone speak with Henrik Hahn, Chief Digital Officer at Evonik, one of the world's leading specialty chemical companies operating in more than 100 countries with over 33,000 employees and 7,000-plus products. Henrik traces his journey from chemical and process engineering through business administration and information economics to leading Evonik's entire digitalization agenda. He explains why AI is better understood as augmented intelligence than artificial intelligence, how Evonik built a cross-organizational "cognitive solutions" agenda using structured use case methodology and OKR goal-setting, why cultural change management is the most important and most underestimated ingredient in any digital transformation, and why - after all the frameworks, governance models, and training programs - there is ultimately no magic. It is just work.
Key Takeaways:
- AI is better framed as augmented intelligence rather than artificial intelligence - the goal is to complement human creativity and decision-making, not replace it, and even the name shapes how employees think about it.
- Highly decentralized organizations should develop a shared framework and methodology rather than forcing a top-down mandate - organizations that do not understand why they are doing something will not sustain it.
- A cross-organizational cognitive solutions agenda that involves both solution providers and potential beneficiaries from day one is more likely to identify the right use cases and sustain organizational buy-in over time.
- OKR methodology - Objectives and Key Results - shifts thinking from output to outcome by articulating a clear objective with quantitatively measurable key results, and enables agile course-correction in cycles before a project dies on the vine.
- Knowledge management systems fail when employees fear losing competitive advantage or personal relevance by contributing their expertise - organizations should ask how many "Coca-Cola formulas" they actually have, because most overestimate how much of their information is truly irreplaceable.
- The most traction in specialty chemicals comes from AI applications in R&D: virtual assistants that help customers find the right additives, internal knowledge graphs giving experts centralized access to accumulated research, patent classifiers, and in silico materials discovery - including a strategic collaboration with the MIT-IBM Watson AI Lab exploring graph modeling, hybrid modeling, and quantum computing.
- Change management is the most important and most underestimated lesson in digital transformation - it is acceptable to be misunderstood for a while, but you must involve skeptics early, listen to their concerns, and let them find the answers themselves rather than surprising them with a finished solution.
Insightful Quotes:
"Artificial intelligence is really meant to augment our human intelligence rather than substitute it. In the first instance you don't have to deal with technology - you have to do everything centered around the human being, the employees, the managers, the leaders within the company." - Henrik Hahn
"As an organization you have to ask yourself: how many Coca-Cola formulas do I actually have? Sometimes we overestimate the importance of a single piece of information and the value of that information. If you can explain that there is much to win but not that much to lose, then sharing becomes the right thing." - Henrik Hahn
"In the end, there is no magic. It's just work." - Henrik Hahn
Tune in to hear Henrik Hahn explain how Evonik built its cognitive solutions agenda from the ground up in one of the world's most complex industrial environments, why cultural science is as important to a CDO as data science, and how a company that makes tires more fuel-efficient, medications more effective, and mRNA vaccine lipids is now using in silico discovery and quantum computing to reinvent how new materials are found.
Links:
Thanks to our sponsors:
Podcast Transcript: Augmented Intelligence, OKRs, and Why Cultural Change Is the Hardest Ingredient in Digital Transformation
Transcript introduction
This transcript captures a conversation between Seth Earley, Chris Featherstone, and Henrik Hahn about what it takes to drive digital transformation and AI adoption inside one of the world's largest specialty chemical companies. Henrik explains how Evonik moved from unrealistic "black box" expectations to a mature cognitive solutions agenda, why he prefers the framing of augmented intelligence over artificial intelligence, how OKR methodology keeps innovation efforts honest and agile, the deep challenges of knowledge management and information gatekeeping in a need-to-know culture, and what lessons he would pass on to any organization tackling large-scale change.
Transcript
Seth Earley: Welcome to today's podcast. I'm Seth Earley.
Chris Featherstone: And I'm Chris Featherstone. Good to be with you.
Seth Earley: Today's guest is a chemical engineer by training who is now the Chief Digital Officer at specialty chemical giant Evonik. His interests include AI, digital transformation, innovation, and strategic change management. When he is not busy with all of that, he renovates classic cars. Please welcome Henrik Hahn. Thanks for joining us today.
Henrik Hahn: Thank you very much. It is really a privilege to join this podcast.
Seth Earley: Tell us a little bit about the car hobby.
Henrik Hahn: I am very much into classic cars - German brands from the 1960s and 1970s. Of course those brands always trace back to famous American producers like GM, so Opel as a former GM brand, for example. I have quite a collection, and I also have some American muscle car models as well as some Japanese cars from the 1970s. It is definitely not an investment. It is for pleasure.
Seth Earley: It is a labor of love. And Chris, I know you renovate motorcycles.
Chris Featherstone: Yeah - I take different style motorcycles and make them into scramblers, cafe racers, or bobbers. I have an affinity toward 1950s style hot rods, blacked out with a pop of color and white walls. And that is when you learn how good you really are - whether you are good with braking systems, or with the engine, or textiles and interiors, or paint. It is very rare that you are good at everything.
Henrik Hahn: I would not say I am particularly good at any one subject, because in the end you also have to know when to rely on expertise. In former days when I had the time, I did not have the money. Now I may have some money but not the time. So you have to source out certain topics. But I have to admit - I also own a BMW motorbike. You really got to me with what you said. It is amazing, but to be honest it is very hard to really be good at it if you do not practice on a daily basis.
Seth Earley: Tell us a bit about how you got to where you are - you started as a chemical engineer and moved into this CDO role - and tell us a little bit about Evonik as well.
Henrik Hahn: I have spent more or less my entire professional career within Evonik and its predecessor companies. I always worked at the interface between technology and innovation on the one hand and strategic business development on the other. For whatever reason, that more and more brought me closer into digital topics. I am indeed a chemical and process engineer by first education, but I also hold a degree in business administration, specializing in information economics and mathematical game theory - so at least it is close to the digital world.
Being part of our strategy group, I proposed that we needed to do something about digitalization - what has been coined in Germany as "Industry 4.0," embracing the entire spectrum of digital technologies. With my technology and innovation background, my former CEO said: Henrik, you talk about it, now go and do something about it.
Evonik is one of the world's leading specialty chemical companies. We are active in more than 100 countries worldwide and have over 7,000 products. Our specialty chemicals, for example, make tires more fuel-efficient, make mattresses more elastic, make medications more effective, make animal feed healthier - just to mention a few. We often contribute only a small amount of material, but those contributions really make the difference. Lipids for mRNA-based vaccines are a very recent example. And by the way - a world without chemistry would be cold, dark, and simply disastrous. No iPhone, no LED lighting, no classic car restoration, and no podcast.
The chemical industry is the industry of industries. And as all industries are transforming, our industry has to transform as well. To keep it simple: we just make stuff for stuff. But when the others who make the end-consumer stuff are changing, we also have to change. Within our process technology group we always try to make our production processes more efficient and more sustainable, which means rethinking processes that have been established over decades. The industry has been pretty successful as a whole, so the particular challenge is: why should anyone change anything? That is also the beauty of my current job - thinking about how one could do things differently than before.
Seth Earley: With 33,000 employees and that scope, the cultural issues around AI often trump the technology and even the data. You mentioned you frame AI as augmented intelligence rather than artificial intelligence. Talk about that.
Henrik Hahn: We all know the famous Peter Drucker statement that culture eats strategy for breakfast. There is much truth in it. And indeed, when we talk about a technology topic like artificial intelligence, even how we call it might change the way we think about it. For me, artificial intelligence is really meant to augment our human intelligence rather than substitute it. In order to do so, you do not in the first instance have to deal with technology. You have to do everything centered around the human being - the employees, the managers, the leaders - to explain what the benefits are, what the threats are, and what the real risks actually are.
AI can really help companies from traditional industries improve the bottom line. This holds true across all key functional areas - procurement, operations, marketing and sales, R&D, finance. And when I joined the company, we were already applying the same mathematical methods that are now rebranded as AI. It is a rebranding, and for good measure. We prefer to refer to these as "cognitive solutions" - they transform data into insights and predictions, and those insights and predictions augment human decision making in a targeted manner.
Chris Featherstone: A lot of people get spun up on the idea that AI is a silver bullet. They miss all the work that goes into it upfront to get the desired outcome.
Henrik Hahn: Absolutely. We have to complement human creativity. That is the key ingredient. And when I was referring to bottom-line effects, of course there is also top-line growth - unlocking new revenue streams. In a traditional industry that is not easy, but it is clearly a must, and human creativity is the only way to get there.
Seth Earley: How are you determining priorities across all those business units? How are you fostering innovation and deciding where to focus?
Henrik Hahn: That is very hard to answer. Take our company - we are a super decentralized company. One could think we clearly need a purely centralized approach to tackle these challenges. I see it completely differently. I believe it is smarter to develop a framework and methodology. That is much more helpful than trying to force the organization to apply something they do not even understand.
To leverage a framework you need to follow up with a cross-organizational approach. We set up a cross-organizational team and a cognitive solutions agenda, and from the very start we involved not just colleagues who can be regarded as solution providers but also the potential beneficiaries. Then we tried to derive a use case methodology. A use case is not a candle you burn down - it is really meant to be there for a long time. Then you evaluate feasibility, applying cross-organizational expertise, so you are more or less confident that you are not just doing things right but doing the right things.
Once you decide on the right thing, you also have to ask: is this something we can do entirely on our own, or does it make sense to involve external partners - service providers, solution providers, technology companies, startups, giants, academia? We have limited resources and have to be very clear about where to do things ourselves and where to rely on an ecosystem approach. And ecosystem approach means the organization first has to learn that it is part of an ecosystem.
Seth Earley: How do you measure success? We have talked before about data-driven decision making and the challenge of having well-instrumented baselines.
Henrik Hahn: We apply the OKR methodology - Objectives and Key Results. It helps not just to set the course but to check whether we are on the right course. You can nail it down to two simple questions: where do I want to go, and how do I know I am getting there? The data answers the second question.
OKR is a methodology coming from Intel - it is also well known from Google. It changes the thinking from output to outcome by articulating a clear objective and key results that can be measured in a more or less quantitative way. It works in cycles, which makes it agile. It is a relatively simple methodology, of course it needs discipline, and formulating those key results needs some practical experience. But I believe everyone can use it - there are great videos on YouTube showing how Google Ventures applies it.
Chris Featherstone: For listeners: OKRs stand for Objectives and Key Results. There is a great book called "Measure What Matters." The key is making sure everyone's fingerprints are all over the objectives - socialized properly, so you can go back and say: this is what we agreed on, this is what it is enabling us to learn and grow from. And if those measures are not mapping to the objective, you course-correct quickly. If you do not have a process to course-correct fast, you will never learn, because you need to fail fast.
Seth Earley: Talk about knowledge management and organizational design. How do you get the right information to the right people at the right time?
Henrik Hahn: I think you are referring to what I call organizational design - everything to do with the knowledge assets an organization has at its fingertips. In our industry we have very much followed the need-to-know principle, which means it was never intended that everybody knows everything. That does not entirely make sense, but bringing knowledge assets to the people who can exploit them is where novel digital technologies can help - intelligent databases, machine learning algorithms to sort and organize knowledge assets.
Most knowledge is still on paper. The first issue was to digitize it - not just digitalize it, but really digitize it. We are talking about enormous amounts of unstructured data. To exploit a knowledge asset you first have to shape it, free the knowledge from where it is hidden. It is sometimes a matter of consistency and sometimes a matter of availability.
And once again it is not that much a matter of technology. It is a matter of trust. It is a matter of people and how willing they are to share knowledge and share information. Think of an organization as an information processing system. It is all about moving the right information to the right people at the right time.
When I first joined the company everyone was talking about knowledge management systems, but those systems were super static. And sometimes employees hesitated - is it really a good idea to put my individual expertise into a system? What about my reputation? What about my relevance? The fear of loss of control was a major roadblock, and those knowledge management systems never really took off.
Chris Featherstone: Do you still find information gatekeepers in your organization?
Henrik Hahn: This is a law of nature. And when we talk about information, we have to be aware that informal information is sometimes very important too. There have been good reasons to follow the need-to-know principle, and we have to deal with these old habits. We have to listen to skeptics, because the skeptics may have a valid reason. We may lose competitive advantage by letting certain information flow. So I need to be clear about where we have to be strict and where we can generate more insights by sharing. We have to categorize and classify information, and be aware that sometimes bringing together different pieces of information changes the classification of what was previously considered sensitive.
As an organization you have to ask yourself: how many Coca-Cola formulas do I actually have? I think sometimes we overestimate the importance of a single piece of information and the value of that information. This needs facilitation, good data stewards and custodians, and it is again a matter of change management. If you can explain that there is much to win but not that much to lose, then it is maybe the right thing. And you have to listen to the skeptics - because the skeptic may have a legitimate reason.
Seth Earley: Centralized standards and policies - taxonomy, metadata, reference architectures - those become corporate assets. But execution has to be distributed. So what AI applications are getting the best traction across your organization?
Henrik Hahn: For our industry the R&D space is really knowledge management par excellence. In R&D it is not so much about efficiency gains - it is really about being at the forefront of innovation, coming up with novel ideas. This is where AI can help develop new materials, modernize processes in light of sustainability, and better understand customer needs.
Some use cases that are already in production: we have a language assistant for the paint industry - a virtual assistant that makes it easier for our customers to find the right additives. We have an internal knowledge graph that gives our internal experts centralized access to the know-how accumulated through our in-house research. We have compound formulations tools, and a patent classifier that is really a practical, useful tool. And we have a strategic collaboration with the MIT-IBM Watson AI Lab where we are looking into graph modeling, hybrid modeling, and even the link to quantum computing - exploring applications in materials discovery. Not classic lab discovery, but in silico discovery. That is a super hot topic, because the current computing power can really be put to work on pattern recognition at a scale that was not previously possible.
Seth Earley: What are the lessons learned for organizations going down this path?
Henrik Hahn: We are sometimes talking about extremely complicated stuff, and I would say it is absolutely okay to be misunderstood for a while. But you have to be aware of the resistance to change, even when you have a good idea. Change is stressful, and people try to avoid it. If we talk about exploiting the knowledge assets a company already has, you will find many people who do not like the idea because it means changing their habits, changing how they do their jobs. In order to avoid the pain, anguish, frustration, and lack of confidence that goes along with change, effective change management is absolutely the most important learning.
This also holds true for any make-or-buy decision - our own engineers and data scientists, of course, believe they know much better than any external expert. That is quite human nature. So you have to involve the right people, involve them early, and do not try to surprise people with a great idea. Educate them, involve them, listen to them, and let them find the right answers themselves.
And in the end - there is no magic. It is just work.
Seth Earley: I love it. The blocking and tackling. There is no magic - it is work. That is a wonderful summation. Henrik, thank you so much for your time today. Wonderful lessons learned about cultural issues, building use cases, measuring results with OKRs, decentralized execution, and foundational education for both newcomers and pros.
Henrik Hahn: Thank you so much, it was really fun. Stay safe, everyone.
Chris Featherstone: Henrik, echo Seth's sentiments - thank you for spending time with us, especially being across the pond with it being late in your day. Very much appreciated. Good luck with the cultural sciences studies and good luck with the car renovations too. Thanks so much.