Ep 66: Productizing AI and Internal Copilots with Eastman CIO Aldo Noseda

Evan Reiser: Hi there, and welcome to Enterprise AI Innovators, a show where top technology executives share how AI is transforming the enterprise. In each episode, guests uncover the real-world applications of AI, from improving products and optimizing operations to redefining the customer experience. I'm Evan Reiser, the founder and CEO of Abnormal AI.

Saam Motamedi: And I'm Saam Motamedi, a general partner at Greylock Partners. 

Evan: Today on the show, we're talking with Aldo Noseda, CIO at Eastman. Eastman is a roughly $10 billion specialty chemicals company with more than 100 years of history. Their perspective on AI is especially valuable because they're applying it across diverse functions while navigating the unique safety considerations of the chemical industry. A few things stuck with me from this conversation. 

First, Eastman isn't only using AI internally, they're turning it into a product. They launched Fluid Genius, which uses AI to predict when thermal fluid in a customer's plant will degrade, so customers know exactly when to act. It's not AI for AI's sake. It's AI packaged as a customer-facing service. 

Second, they've taken an AI-for-the-masses approach with an internal engine built on industry solutions like ChatGPT, wrapped with security and company customization. The numbers are kind of wild: 6,000 recurring users, programmers going from 5,000 lines of code a month to 40,000 with AI agents, and an AI-powered IT help desk stood up in two weeks. 

And finally, Aldo made a point I think a lot of leaders need to hear. Risk tolerance has to be situational. If you're building a coaching tool for sales, you can take some risk. But if you're opening or closing a valve in a manufacturing plant, you better have the right answer.

Aldo, first of all, thank you so much for joining us today. Maybe to start, can you give our audience a bit of background on your career and your current role at Eastman? And maybe for people that aren't familiar with Eastman, give a little context. I think probably people don't quite understand the impact and the scale of operations you guys run there.

Aldo Noseda: First of all, thank you, Evan, very much for inviting me to this podcast. Aldo Noseda, CIO at Eastman. I've been with Eastman around eight years, coming from a very long career with Monsanto of 27 years before that I started my career there with Monsanto. I had multiple roles in supply chain and commercial as part of my career, but always coming back to IT. About Eastman, it's approximately a $10 billion company focused on both commodity chemicals as well as specialty chemicals. A company of more than 100 years, trying to make a difference in the world. In the last years, we have been putting particular attention to this idea of circular economy as an add-on to our specialty set of products.

Evan: Maybe to start, I'd love to hear how you guys think about approaching AI development, especially in your organization that's so multifaceted and so diverse in terms of different parts of the business. What role is AI playing for you guys today?

Aldo: First of all, you're right. There's a lot of moving parts on the AI journey. 

We had the fortune that we created a data science team many years ago, so we had the foundation, and we started with statistics, machine learning, operational research, and artificial intelligence. It was in the last couple of years that we really started to pay attention to generative AI. The way that we are approaching the problem, first of all, before we go there, is we don't know how much we're living through a historical moment or how much we're living through a hype but we are assuming that it's going to be important, that the impact is going to be really, really impactful. Because of that, we are tackling both the yin and the yang. We're tackling how we can bring capabilities to our company to make a difference, as well as mitigating risks, and we're treating both sides of the coin very carefully because we believe both are important. 

But whatever our plans today are, they're probably going to be different a year from now because this is moving fast and changing, and we're trying to be dynamic in this perspective.

Evan: Are there any particular innovations where AI has augmented some of the conventional research processes that you feel really proud of, or what's had the biggest impact?

Aldo: What we are doing right now, we are experimenting with this model of trying to read the tech reports that we have historically in our archives and then try to bring combinations and knowledge that we didn't have before. It took a lot of time to pull it together and try to accelerate. 

I think we're going to see in the next year the impact of what generative AI, on top of what we have done with data science, can do to R&D. But obviously we are trying to apply the same concepts to many areas of the company beyond R&D. R&D is just one of the examples. 

The big difference is, before, like we said, in R&D specifically, machine learning, statistics, and operational research were all techniques that help discover products, but in general you needed structured data for those models to work.

Evan: Are there any use cases that are maybe outside R&D? If there's another CIO listening and saying, "Hey, I want to steal all those pro tips about smart things with AI," what about outside R&D?

Aldo: We at Eastman, in the last years, started to do something fairly unique for the chemical industry, which is that we started to offer to our customers digital solutions in the form of services. 

We have four products in the market right now that we are offering. One of those products is called Fluid Genius. What is Fluid Genius? We sell something called thermal fluid. It's a liquid that you use in your plants in order to keep heat around the plant without the need to put multiple furnaces across your operation. Our customers that own that operation don't want that liquid to degrade over time because they want the heat. The problem is that this particular liquid, which is the thermal fluid, has degradation once it starts. So we were able to develop an AI capability that predicts the degradation of the liquid over time. Now we have embedded that engine inside Fluid Genius, which is the name of the product, and our customers can put in some data, put in some sampling information, and get a prediction from the engine of when their liquid is going to degrade. Why is that important? Because then they know when they need to stop or plan some maintenance for the plant, which of course translates into money. Our customers love it. We are having a lot of success. It's another example where, again, artificial intelligence and data science are starting to make a difference. And I could go on and on. Those are two good examples, I guess.

Evan: So am I understanding correctly where the AI will help, there are sensors that will sample the thermal fluid and basically give a preventative maintenance warning so they can prepare for maintenance? It's not unscheduled.

Aldo: Absolutely. But the difference here is, I think you're absolutely right. I think the thing is, certainly we work in data science and artificial intelligence as much as possible for the maintenance of our plants. This has been a niche offering for our customers in order to help them with the maintenance of their own plants. But you're right, and conceptually that is what it is, particularly targeted for the product that we sell, which creates a very interesting combination.

Evan: What would be your advice to another CIO out there about, hey, this one's easy, just if you do this, it's a quick win, it's not that hard, and it's going to be a crowd-pleaser? What would be your advice about one more thing to hit the ground running and get more AI transformation?

Aldo: We thought about how can we provide capability for the individual across the organization without the need of a complex, heavy IT dependency. There were tools that were appearing in the market, the ChatGPTs, the Geminis, the Microsoft Copilots. So what we had to do was create an engine utilizing, obviously, the base of the existing products in the market, but wrap it up with a solution that was not only secure, but customized to the needs of the company, and we deployed that very quickly. Right now we have approximately 6,000 recurrent users utilizing that engine for individual consumption. 

That is what I call AI for the masses. 

Then the more sophisticated use cases, we are tackling two things: one, advanced use cases for growth or opportunity, where we're using, again, advanced mathematics and generative AI for solutions like, for example, the sales one that we've evolved, but also agentic AI, where we can complement the work and the repetitive, boring work of some of the individuals, and we're tackling the two at the same time. You start with the small things. 

As an example, we had one example that's very well known in the industry. In information technology, we have a help desk function. The Tier 1 demand of the help desk is not a very interesting activity to do by a person. It's recurrent. It's always the same. You just read a script and you have it. We loaded the script, we put it on top of the help desk, and in two weeks we had the engine up and running for our users to consume. Even in the IT organization, we are seeing phenomenal progress in our performance around coding. Someone told me we went from 5,000 lines of code per month per programmer to like 40,000 lines of code using AI agents, which creates a phenomenal acceleration of software to help our business.

Evan: A couple weeks ago, I talked to Adam, your CISO, about how they do things in security. He was talking about, "Oh, this one project was jumped on the whiteboard. We kind of all work together." How do you think about setting up the culture of the technology team more broadly to help drive some innovation?

Aldo: I have this theme about staying humble but hungry, the double H. You probably heard it before. I push a lot to this: let's be very respectful, a little collaborative, but at the same time, let's go after it. Let's push for it. I also share with my team sometimes, do you know how Italians cook spaghetti? They say no. Well, you throw it to the wall, and if it sticks, it's cooked. Well, that applies sometimes to technology as well. We don't need to be perfect the first time. We just need to try things and see how it works and then push it forward. I think from a business perspective, especially with this generative AI trend, people at the beginning thought, hey, there's OpenAI, there's ChatGPT, there's Gemini. They're like big Google search engines, right? 

And that's what people thought at the beginning. Then they started to realize, oh my God, this is much more powerful than that. I think the biggest challenge that we're going to have is how the organization embraces the power of what appears to be coming and how the IT organizations can help ease that path for success that translates ultimately to value.

Evan: What about maybe on the physical side? Maybe we're not there yet, but you can imagine the future use of AI to control robots or machines, especially in an organization like Eastman, where you have a 100-plus-year history of safety culture that requires a different level of trust. In some ways, that trust from IT leading this initiative is kind of built in the knowledge-working world, but eventually will go in other places. Continue on the cultural theme. What are some of the ways you work on building that trust?

Aldo: This gets back to the risk scenario, right? You need to be situational. We also are very careful. These models, in essence, are probabilistic models. They can hallucinate. They can create outcomes that may not be correct based on the data that they are consuming and the time of processing that they have, and they could create errors. Well, if you're creating a coaching opportunity for the sales organization, I think you can take some risk. If you are opening or closing a valve in the manufacturing plant, you better have the answer right. I think we're going to need to be careful where we deploy these models and the level of scrutiny that we have in the environment. So if we are situational with the risk tolerance and how we think about the different aspects, I think we're going to be okay. But these capabilities, if we believe that this is a historical moment and less of a hype, are going to become more powerful. And as they become more powerful, we're going to need to also become more careful about where we have outputs that could influence or change something that we don't want to change.

Evan: What do you think is the most important part about AI that's being a little bit under-discussed, maybe across the media and across your peers? What do you think we need to talk more about as an industry?

Aldo: On the hard side, I think there's a lot of energy around the possibilities, and I think the world is missing a little bit of this conversation that we talked about, of the translation-to-value part of the equation. On the soft side, which is the change management, there's an assumption that everyone is going to jump into the wagon, and I think more discussion about psychology and human behavior is probably going to become important. So it's again the hard and the soft. In both, I think this is going so fast and the discussions are so quick that maybe a little bit more on those sides could be convenient, but it's what comes to the top of my mind on those ones.

Evan: Do you have any advice for your peers out there, things that have worked for you or ways that people should be thinking about the problem a little bit differently?

Aldo: If you think about the adoption of the organization that's trying to embrace AI and how we're doing, we launched a very strong training program. That was the first thing that we thought about. It was combined with a communication plan, combined with a change management prep plan. But then it's also a little bit the show-me story. I think as you get people embedded and trying things in a safe sandbox and they start to create more and more outcomes, I think there's going to be an acceleration of utilization and opportunities. Also, I do believe in the guardrails to prevent a risk problem, but I also believe in the decentralization of the AI capabilities in the organization in order to provide scale and acceleration with the help of everyone. All those thoughts on how to unlock the potential, knowing that there's going to be a change curve and an adoption curve that will need to occur.

Evan: What would be your advice to someone trying to follow in your footsteps and wanting to play a bigger role? They want to help transform organizations. What would be your coaching or advice or mentorship for them?

Aldo: I think it transcends AI. It gets to any career, particularly information technology. We talked about the hard and the soft. I think that IT professionals in general are very logical, are very well structured, have a very abstract and tactical view of things. In many cases, not in everything, the communication skills are not very polished. You talk about the IT nerds and how they work in their own thing. Well, not everyone is like this. Let me clarify that. But there is a need to develop technology individuals that are also very good communicators, and now it's going to become even more important. 

The problem that I see happening is that the kids in their careers, especially in the IT career or computer science or whatever you call it, analytics, they are getting recognized by their logical results and performance, and not many times by their communication skills. 

But one day those same individuals become leaders, and not that they don't know how to talk, but they're not great communicators to influence, to sell, to convince. So I always tell junior people in the organization, push yourself to develop your communication skills. Go and take a class in theater, loosen up, and do something of that nature, because it's going to be invaluable in your career to develop those skills that normally at the beginning of your career are not being trained. 

In general, the DNA of the technologist is not conducive to the communication angle. It varies by person, of course.

Evan: When you think about your leadership team, what are some of the things that you're starting to value more than maybe two years ago, and less?

Aldo: I think even at my leadership team, or the next level of the leadership team, I was in a town hall yesterday, actually, and there was a very strong leader that was talking about her career, and she said, "My career grew, and 10% of my opportunity came because of my performance, 90% because of my relationships and my network." Wow. How important is talking and communicating and interacting? So that for me was a little bit of shock. I don't know if it's 10/90, but there's a weight that many times we don't think about. And I tell my leaders, especially those that are trying to influence and push for change in an evolving world, you need to be out there. You need to be out there selling and convincing and influencing and challenging and all those good old things. And now, to your point on AI, there's also this aspect of interacting with natural language and visuals in an environment that again goes less to the hard and more to the soft. I think there's also going to be a play even for the more junior people who are going to be using these technologies. Maybe we'll see.

Evan: Aldo, what we like to do at the end of the show is a bit of a lightning round. We basically ask you four or five questions that are very hard to answer in the one-tweet format. We're looking for a shorter answer. I'm going to kick it off with a couple questions, and forgive me in advance for making these hard to answer succinctly. Okay, so question one: What is the best way for a CIO to stay up to date on current enterprise AI trends? Any advice for an incoming CIO about how they stay up to date with the latest technology, given AI is changing every hour, every day?

Aldo: Networking, either with your peers or with technology companies that are evolving in the market, is key. There's a lot of information in papers, but chatting with people that understand this thing makes a difference.

Evan: For this next one, it doesn't have to be a work thing or an AI thing, but is there a book you've read at some point in your career that's had a big impact on you? If so, I'd love to hear what it is and why.

Aldo: I always go back to the traditional Seven Habits book from [Stephen] Covey. It has some fundamentals that now, when you read it, you say, "Okay, table stakes," but it's fairly good, at least to organize your life and organize your thoughts and the way you're thinking.

Evan: What's an upcoming technology you're most excited about? Something you're paying attention to personally?

Aldo: I am just trying to learn here about the evolution of quantum computing because I believe this combination of software from an AI perspective, now expanding to the use of voice or visuals, combined with power connected to quantum, I don't know where it's going to go. People are saying 10 years from now. I don't know, but it's a very interesting thing that I'm reading.

Evan: Do you think in the future all your software would be personalized for Eastman, or do you actually want it to be more platformized, so you're taking some of the best practices from some of your peers in the industry? Where do you think we'll end on that spectrum?

Aldo: If you remember, we talked about this person that was doing 5,000 lines of code and now 40,000 lines of code. So they are developing software. A capability that took us two months to build, we just built it in 10 minutes the other day. So this is real. The question becomes how sustainable are those solutions going to be, and what are the quality levels, how enterprise-ready are they going to be? I think there's going to be a transition, to answer your question, where this would go, or how much is going to be a bunch of agents developing software and the traditional packages start to fall apart, and platforms to build are going to become more important. 

Evan, I don't have a clue, to be honest, but I see the trends coming, and I ask my team, let's experiment. Let's start to build capabilities using AI agents. But if there's a critical piece of enterprise software, don't jump on it too quickly. We need to also make sure that they're sustainable and high quality. So again, being situational on these decisions while all this evolves is, in my opinion, the wise thing to do.

Evan: Well, I appreciate you joining the show, Aldo. Hopefully we'll get a chance to speak again soon, and thank you so much for joining us.

Aldo: Hey, thank you very much for everything, and I'm glad that it was a really good conversation. Thank you very much.

Evan: That was Aldo Noseda, CIO at Eastman.

Saam: Thanks for listening to Enterprise AI Innovators. I'm Saam Motamedi, a general partner at Greylock Partners.

Evan: And I'm Evan Reiser, the founder and CEO of Abnormal AI. Please be sure to subscribe so you never miss an episode. Learn more about enterprise AI transformation at enterprise software.blog. This show is produced by Abnormal Studios. We'll see you next time.