Crafting User-Focused Solutions and Building an AI-Driven Culture

In a recent CDO magazine interview, Sue Pittacora, Executive Advisor at Wavicle Data Solutions, and Abhi Seth, the Chief Enterprise AI and Data Officer and Vice President at The Boeing Company, sat down for an in-depth discussion about how Boeing leverages AI and data to drive enterprise-wide transformation, prioritizing quality, safety, and workforce empowerment. 

Throughout this discussion, you can learn how to embed AI into operations to solve complex challenges and optimize processes. 

Watch the full interview here or scroll down for a detailed transcript of Abhi’s insights. 

Sue: Hello and welcome to the CDO Magazine interview series. I am Sue Pittacora with Wavicle Data Solutions, and I’m delighted to be here today with Abhi Seth, the Chief Enterprise AI and Data Officer and Vice President at The Boeing Company. Abhi, thank you for joining me today.   

Abhi: My pleasure. Thank you for having me.

Sue: Today, we’re going to explore how Boeing leverages AI and data to drive enterprise-wide data transformation, prioritizing quality, safety, and enabling people. So, Abhi, let’s just jump right in.

What is Boeing’s AI strategy and how does it reflect in the company’s technological innovation?  

Abhi: Thank you for the question. We have a tremendous history of driving technology innovation and AI, and AI is not a new thing at Boeing. We’ve been applying AI in many forms in product development and customer enablement for decades. With the recent evolution of areas such as high-performance computing and generative AI, AI has become more of a buzz and is becoming a lot more active. We have a robust AI strategy, and a lot of the focus of that AI strategy is driving sustainable competitive advantage for Boeing.  

Our strategy focuses on four key pillars, and our AI strategy’s cornerstone is the data foundation. Without data, you really can’t go too far in driving your digital or AI strategies forward. So, how do we master our data?  

Then the second is around platforms and infrastructure. How do you scale AI? Many of us want environments where we can experiment and prove our different concepts. But how do you quickly scale to create that impact on a large scale? So, your infrastructure becomes a key cornerstone of that.  

Then, we look at building the right talent and upskilling the capability across all our practitioners on awareness of the art of the possible with AI, how to identify the right problems, and then building a digital and AI-savvy workforce across the company. So, talent becomes the third pillar. 

The last one is bringing domain, data, and AI capabilities together to create more differentiated business capabilities that move our strategy forward. So, we are looking at four key product family areas. The first one, of course, is safety and quality; how do we drive domain data and AI expertise to create capabilities around safety and quality to drive enhancements? Then the second distinctive area is engineering and manufacturing, the third is supply chain, and the fourth is customer success in the aftermarket. So those are kind of the four key areas where we are focused on building AI capabilities.  

How is AI used at Boeing across manufacturing, supply chain, and customer success?

Abhi: A lot of focus at Boeing is really around safety and quality. One of the things I want to highlight is that we have had a lot of quality standards in terms of taking a day to discuss how we can improve quality and safety around our factories and in the manufacturing space. We have gotten a lot of feedback from our fellow mechanics. I would like to give you examples of a couple of things.  

So how do we drive AI and generative AI capabilities to condense and synthesize the feedback we are getting? How do we summarize the common themes and pain points and be able to prioritize some of those pain points using AI? The second and probably more pertinent one is one of the things that came up with our mechanics. They were having a hard time reading our drawings on the shop floor. So, as they were trying to apply work instructions and read a drawing on how to perform a certain assembly operation in the factory. Some of our drawings in certain areas were not as clear, and we were able to apply computer vision to darken those drawings.   

The previous process was really around how to send a drawing back, physically darken it, bring it back, and re-scan it. Now we can apply computer vision in real time to darken those drawings and make readability easy. So now this is released to all mechanics. It’s part of a production system just within three or four months, and that capability as you’re trying to read a drawing, if there are certain things you want to enhance or you want better readability of the drawing, you’re able to use that computer vision model in real time in some of our applications.   

The other thing we’re doing around quality is really how do we get visibility of quality across our site, and not just within Boeing sites but also from our suppliers? So, we were able to integrate quality data not just from our factories but also from our suppliers so that we can drive real-time visibility of quality metrics and insight for our production workers.   

What challenges have you faced while implementing AI at scale at Boeing? 

Abhi: If you think about driving digital and AI, a big chunk of that is driving the digital literacy of such a large company. How do we make sure that everybody is getting up-skilled? Also, when we’re deploying some of these solutions to make an impact, one of the things we need to do is solve the right problem and make sure the solutions we are building are easy to adopt and use by our end users. I think that’s where many challenges happen in terms of change management, adoption, and sustaining some of these capabilities while they are being used.   

So, we start with the business problems instead of saying we will do AI in this space. We are trying to get to the forefront of what problem we are trying to solve, the personas who have the problem, and getting those personas involved very early in our design-thinking process on the solution.   

I think what is undervalued in a lot of these applications is how important it is to double-click and truly understand the problem statement itself. So, these are things that we emphasize in our operating model: spending time understanding the problem statement, talking to the end users, coming up with the solution with them, involving them as we are building some of these capabilities, and then doing a small pilot. A lot of this is iterative and feedback-driven. When you do this user involvement and understanding of the personas early, it makes the solution much easier to adopt. We get the right solution the first time versus having to build something in isolation and deploy it, and nine out of 10 times, it becomes a challenge. It’s like, hey, we didn’t get something right.   

The other thing we look at is that I think a lot of the focus on AI capabilities is augmenting and enabling our workforce. Enabling the mundane tasks that people sometimes need to do. So, it’s about involving them and understanding their current workflow. Along with the AI capabilities, think about what would their workflow be, and how would it unlock productivity? And not just productivity but think about it as, “how do we make the right way the easy way?” And, “how do we make everybody’s job easier?” And, “are we making it simpler?” So there is no change management, in effect, involved because they love doing it and now it’s much simpler. So, that’s kind of how we have approached it.   

In terms of challenges, I would say that in the overall AI space, one of the biggest problems I see is whether we are really solving the right problem itself. We have so many opportunities in this space we could be solving. If you look at manufacturing or supply chain, these spaces are very big, and we have a limited capacity to execute. A lot of times, we are looking for problems that are in the intersection of how to align with strategic priorities of the day, what is the most important thing we need to solve, what is the feasibility of solving it, and what is the size of the impact. So, if none of these three things align, then it is not the right problem to solve most of the time. That’s how we pick the right problem. I think it’s training the community overall to figure out how to pick the right problem.  

Then, in solving it, I think a lot of the focus we have done is how we segment the problem. Some of these problems are too big. They have been around for a long time, and I think just applying some AI pixie dust will not solve it.   

So, thinking about problem decomposition, how do you decompose the problem into small problems, and then how do you solve a meaningful portion of that problem and deploy a solution and see how it works and by doing that, really gain an understanding of that subject matter or that domain? A lot of the focus we have is on domain teams, which is in AI, so that as these teams continue to work in manufacturing, supply chain, or operations, they are gaining domain expertise in those areas over a period of time, building the relationships, understanding what the challenges in that field look like, and what does the data look like in those areas.  

Hence, they can come up with better solutions overall. This way, they can create a more realistic roadmap view. And this is not like building point solutions. We are saying there is a problem; let us develop this solution, deploy it, and see how it works. Let’s build upon this as version two and then build upon that. Then suddenly, within three-to-six months, we have multiple solutions that are building upon the whole suite. That whole functional area is starting to get transformed over time, and that’s kind of our approach.   

Can you share some examples of specific AI-driven initiatives that you’ve been able to leverage at Boeing to impact the business? 

Abhi: Absolutely. I think I talked about this example around safety and quality, where we have deployed this computer vision tool to help our mechanics leverage our drawings better.   

I’ll talk about a few others. A lot of our work at Boeing involves writing proposals and responding to customers’ requests for information or proposals. I think that as an older company, we have an opportunity to simplify processes and digitize them, improve the cycle time, and drive more throughput.   

So, we are looking at our response time process. We have around a thousand procedures, for example, for how to respond back, how to write a proposal, or how to do this right because we are a very process-oriented company. But we are using generative AI to figure out how to create a co-pilot for a proposal writer. That is an interesting example of us piloting with around 60 to 70 people. We have so many procedures that it’s hard for people to figure out how to find the right procedure. Especially when you’re coming in new, it can be overwhelming. So, we’ve used gen AI, ingested, and created a custom solution, basically a proposal co-writer. To that proposal co-writer, you can ask, “hey, I have this kind of proposal. What are my steps to respond to a proposal? How do I do this?” It’s trained in all our processes and can guide you on how to write the proposal back, which is pretty cool.  

We even did some time studies to determine what is truly beneficial and found that among new users, there is a 90% reduction in time because they are sometimes waiting to figure out where to find this information or waiting for a subject matter expert to be available to guide them, and now they can find it in an instant. 

More interestingly, we did the study on how it is helping experienced people and saw around 50% improvement in their cycle time, because even though they have been here for a while, there just are too many procedures to keep track of, so they’re able to help with that. These are some examples that I would like to highlight. 

Also, on the supply chain side, we are actively working to predict and anticipate supply chain risks and plan around them so that we can drive minimal disruption in our factories.

Sue: Well, Abhi, I think that’s a great way to wrap up and talk about how your culture embraces this. You did a great job today. Thank you so much for all your insights and for joining me today. To all, please visit CDOmagazine.tech for additional interviews.  

Abhi: Thank you.  

This is the first part of a two-part interview with Abhi Seth covering pressing advanced analytics and AI topics. Stay tuned for the release of the second part, coming soon, or read more of our advanced analytics thought leadership here.

Need help to get started on your AI journey? Get in touch with Wavicle’s experts.