onsemi on AI, Agentic Systems, and the Digital Thread in Semiconductors

Chad Watt:

Welcome to the Infosys Knowledge Institute podcast where business leaders share what they've learned on their technology journey. I'm Chad Watt, researcher and writer. Today I'm joined by Ratish Kumar and Christine Deliance from onsemi. Onsemi produces semiconductors for automotive and industrial applications worldwide. Ratish and Christine lead technology across the business from product inception to sales. Welcome to you both.

Christine Deliance:

Thank you.

Chad Watt:

Christine, let's start with a quick overview. Can you describe the types of chips that onsemi produces?

Christine Deliance:

Chad, onsemi specializes in intelligent power and sensing technology. So you don't see our chip, right, but they're everywhere. They power cars, electrical vehicle, that enable them to go further with just a charge. They also help support the AI data centers that fuel our digital economy and also smart manufacturing. So all in all, we have, we support, we serve multiple markets. You mentioned automotive, but also industrial, consumer, medical, and smart buildings.

Chad Watt:

So Ratish, onsemi has grown through acquisitions. How has integrating these different businesses shaped the technology strategy?

Ratish Kumar:

Technology is an evolving field, and acquisitions are a fact of life. And semiconductor as an industry has a lot of acquisitions and divestitures. So part of the strategy is to look at what are called pillar platforms, develop core technology frameworks that can be used or adopted as new acquisitions are made. So in other words, you develop a core framework, whether it is how do we go to market for customers or how do we manage order management, right? And those frameworks are developed once. We call them pillar platforms. And then as acquisitions come in, they grow into that pillar platform. So this way, when we have acquisitions, the time to acquire or time to integrate is way faster.

Chad Watt:

Ratish, can you explain the concept of a digital thread?

Ratish Kumar:

So the concept of a digital thread is how do we streamline and optimize business processes enabled by technology across multiple business functions? So in other words, how do we ensure that marketing is tightly integrated into sales? And how do we ensure that sales is tightly integrated to service? And service generates customer satisfaction and how do we convert that customer satisfaction into more business through marketing and sales. So how do we streamline business processes that goes across these multiple functions? Number one, enabled by technology, which is where the digital in digital threat comes in. That is the concept of digital thread.

Chad Watt:

Christine, tell us about onsemi's approach to artificial intelligence.

Christine Deliance:

Onsemi is looking at AI as really much more than just automation. It is about reimagining our business and looking at AI really as a teammate so that we can enable collaboration between agentic AI agent and actually humans.

Chad Watt:

How is artificial intelligence being adopted across the enterprise from leadership all the way to sales?

Christine Deliance:

So actually that's a great question because we take a two-pronged approach. First, actually we established at the leadership level a governance structure where we have representation from each executive of the various function. And the mission of this board, if you want, is really to evaluate the various AI project across the company and then determine which one we're going to fund, which one we're going to execute. And that enabled us to basically focus on AI projects that are really going to move the needle and generate outcome. The second, I would say, point is really more bottom-up approach. Where basically we're establishing a community within the company of AI users, where people can actually learn how to use AI. They can also exchange ideas on use cases, the type of prompts that has been really useful for them. Basically, this is creating this environment of collaboration among the employees and to increase the level of AI fluency.

Chad Watt:

Great. Sounds like you have kind of a top-down mandate and strategy, but also kind of a bottom-up organic way of sharing and experimenting.

Christine Deliance:

Yeah, absolutely. Actually, what's really interesting is that we are partnering with our HR colleagues from organizational development to help us in, I would say, rolling out this new way of working within the entire employee population. And I would say they have provided invaluable insights.

Chad Watt:

Ratish, can you share an example of how onsemi is using AI to improve operations or customer engagement?

Ratish Kumar:

Yeah, absolutely. I mean, end of the day, we're all about meeting the customer where they are. End of the day, right? We use technology for that. We use AI for that. An example is, from a customer's perspective, he has a question. He has a technical question. Now, in the past, he would either send an email or reach out to customer service. What we've now introduced is an AI agent. We call it the tech support agent. So you as a user can come in and ask a question. The agent, AI agent, gives you an answer. Self-serve, just like we do in our personal lives. Right? Now, we can stop there. What is really interesting is we said, let's make sure that the answers that's been given by the AI agent is good. So we've introduced a human in the loop program, where once the question is answered by the AI agent, that transaction, that interaction is recorded in the system and there is a human, a team member who actually goes in and reviews those answers. If it's good, gives them a thumbs up. If not, we reach out to the customer and say, hey, we said this, but guess what? There's a small change or nuance that I want to clarify and then we provide that insight as well. It does two things. One, the customer now knows that we've got his back. We are able to give him confirmation, or we are checking and we are validating that what he's got is exactly right. Number two, it helps us improve our AI algorithm or AI logic to improve the accuracy of the data that we provide. Third thing is it really elevates the value add of our human resources. We now have our human resources, our team members doing more value added services. You're going from the mundane to more value-added services. So that's how we've implemented this program, and it's had really good feedback so far.

Chad Watt:

Christine, I want to take this back to sales real quick. How is AI being used to increase sales or make cross sales more effective?

Christine Deliance:

Yeah, actually, you know, this is a very promising area for us to leverage AI to drive growth, particularly the space of cross-selling. We recently actually deployed a solution that helps sales to uncover new opportunities for cross-selling where we use actually, we have AI that analyzes the historical ordering pattern that the design when behavior of our customers. So, and additionally also, what I would call reference application design, if you want, that is prepared by our application engineering team. So all of this to create basically a very insightful source of knowledge that can provide meaningful recommendation. However, you mentioned earlier that AI is not only about technology, it's about the people. And I think this is one of the key lessons, right, that we have gotten from this implementation, where basically if you don't integrate AI in the flow of work, then it's a little bit harder. It becomes more like a task. It becomes more like, it does really add all the value that it could be. So this is where we're learning. We're planning to do more focus groups with our engineers to see how we can deliver more value with AI. And then this concept of AI in the loop also is extremely important because in order to refine our model, to have our recommendation to be better, so we rely on our field application engineer to provide feedback so that we can continuously improve.

Chad Watt:

What impact has AI had on customer satisfaction?

Ratish Kumar:

The whole idea of the whole strategy right here is to meet the customer where they are. So how do we handle our business today on a personal life? If you had to place an order online, you want to know what the price is, you want to know what the ratings are, so on and so forth. How do we meet the customer where they are? So from a customer experience standpoint, what we want to do is to make the life of a customer really simple. An example of that is, you want to place an order for an item? Click a button. The agent is an AI agent, is able to help you with that. And where it cannot, there is always a customer service rep who's there to help you. You want to change the data on a shipping data on an order? No problem. Here is, you just talk. Have an easy interaction. Hey, I'd like to know what the status of my order is. Here it is. Can I change that date, please? Yes, you can. Let's do that, please. OK, here it's done. Here's your acknowledgment. Make life simple for the customer. And how we do it internally is our problem, right? But end of the day, how do we make this seamless for the customer? So how do we, like Christine was saying, how do we go from a traditional model where that is a, the customer reaches out to somebody, the somebody then responds, that is a delay in timing, and then go back and forth. How do we make it more interactive? In an easy conversation, just like we are talking, it's just that there's an agent on the other side who's actually helping you with stuff. That leads to customer engagement, it leads to customer attention, it leads to customer loyalty, and it shows in our NPS scores.

Chad Watt:

What onsemi does in making microchips is a pretty complex and established process. Introducing AI into established processes can be challenging. How do you persuade skeptical colleagues to embrace these AI tools?

Christine Deliance:

Yeah, that's actually, I would say a paradigm that my team faces on an ongoing basis. Just like for example, this week we had sales colleague who had question about a newly AI generated executive briefing that we do to prepare executive for customers. And their reaction was, where does this data come from? And I don't think it's right. And so I think it was a great opportunity for us then to engage in a conversation where to explain where the data came from, how the model was using it, and what's really important with AI is the data is at the foundation of everything. So in this case, a lot of the information that was used in this executive briefing was actually coming from the sales people maintaining information in forecasting systems, in opportunity management system. So at the end of the day, if you want to have good recommendation with AI, you need to have good data and you need to have the right processes and you know to enforce the population of that information.

Chad Watt:

I'm going to double click on that because you're talking about data and building trust. What are the strategies you've seen that have worked best for building trust in AI-driven decisions?

Christine Deliance:

So I think it's all about being very open and engaging the user as early as possible so that they can understand how the model is built, that it's not a black box. And they can really start trusting the data on the system. Another thing also that for us is very important is capitalize on quick wins and celebrate those quick wins. I was talking about creating community of AI enthusiasts at onsemi. This is basically to promote what any of those users have done to maybe save some time, et cetera. But it's really to promote those wins. But I think what I would like to point out is that another thing that's really important is to really improve or increase the AI fluency within onsemi. What I mean by that is in order for people to really embrace AI, they actually need to use it themselves. They need to start using pomp and practice. And that's really absolutely critical. And as they develop those skill sets, then they start embedding it in their way of working. And they can influence their colleagues and around them to basically drive this adoption.

Chad Watt:

Ratish, you've talked about AI agents. Can you give me your high-level view? What's your outlook on AI agents in semiconductors?

Ratish Kumar:

I think foundationally, there's three primary areas where AI is going to change the world. We've talked about AI agents meeting the customer where they are. So these are customer-facing agents. That's one. Number two, how do we use AI agents to improve productivity within our own employee base? How do we improve, how do we get our sales guys to do better sales? How do we get our service to do better service? How do we improve, it's a productivity play within our own teams, right? Use AI for our employees. And the third area, and this is an evolving space, is how do we get these AI agents to talk to other AI agents? For example, if we know that there is a supply chain and there are certain areas which cause certain delays in that supply chain. Can we introduce AI agents that can monitor those and proactively prevent those failures from happening? So three foundational areas, customer facing, employee facing, and how do we streamline these processes so that we don't have those issues in the first place. That's the agent on agent approach. So these are the three areas that we're looking at. The foundational concept is how do we use technology to do this? So there's two parts to it. That's the business process streamlining and how do we use technology to enable those business processes with a strong foundation of data and agent on agent collaboration. That is the framework.

Chad Watt:

It's a great framework. Are there any pilot programs or plans to deploy AI agents in onsemi that you can share?

Ratish Kumar:

Yeah, absolutely. We talked about the tech service chatbot, customer-facing chatbot earlier on. There is an initiative to say, how do we streamline those three things that I brought up? How do we introduce agents in the area of commercial service? What we call customer service, we call it commercial service at onsemi. How do we allow for agents to do self-serve to a customer? Number two, how do we use agents to help our customer service reps be more productive? And number three, how do we use agentic AI on AI conversations to prevent those issues from happening together? All of these three concepts are being looked at from a commercial service space first. That's an example of how we're using all three levers of AI in one area of commercial service. And of course, there's quite a few that's down the pipeline. There's a massive list, like Christine was saying. We have a whole mega list of25 things we want to do, but we're going for things that impact the customer first. And this is one idea, customer service, for sure.

Chad Watt:

Thank you guys so much for your time today. Before we close, is there any question I haven't asked that you guys want to address?

Christine Deliance:

I think what is important in this world of AI where things change extremely fast is to create a culture and an environment where people can experiment and they can try, potentially fail, but then adapt and adjust because as you know the old adage, the quicker actually you fail, the less expensive and you can move to the next challenge. Something also that I would like to point out is learning, this ability of learning very quickly is really a competitive advantage. Often, particularly for example in sales, people fear that AI is going to take their job, that basically AI is going to replace a salesperson. But actually, that's not really what's going to happen really in the short term. In the short term, it's going to be salespeople who actually learn how to use AI to their best benefit, right? That will replace salespeople who don't know how to use AI. I mean, for me, I think this is important. Last few days, right, I was at Dreamforce. And the Chief People Officer at Dreamforce was doing a session on Dreamforce. And one of the things she mentioned is exactly along those lines, which is that learning is a new meta skills. And we absolutely need to infuse that culture, that enthusiasm, if you want, to our employees across the company.

Chad Watt:

An interesting way to think about learning as a mindset but also a skill and a practice.

Ratish Kumar:

So another thing that I would like to point out is how do we measure the ROI of those efforts? Because that's really where the rubber hits the road, right? At the end of the day, to deploy all those tools like Copilot and agent, et cetera, this is great. But we talked about it earlier, they need to be adopted. I think foundationally, AI is here to stay. So the faster we get on it, learn, make mistakes, learn, improve. Do that and use technology. Technology is a friend. Technology is foundational. Technology is an enabler. Use the technology for your benefit. That's number one. Number two, create, have a strategy in place. One thing I've seen is, and I've talked to many colleagues from other companies, everybody's doing its bits and pieces all over the place. Have a governance council. Have a top-down view of what is important to the company. Where the most bang for the buck is. Identify your top three use cases and go in, and go all in, right? And when you decide to go all in, point number three is have a clear view of the framework. Have a digital thread. Know where your data is. Know how to connect the data across these functions. Because it's garbage in, garbage out, right? AI will be successful when you have the foundational digital threads in place and the foundational data in place. And then AI is the cherry on the cake. So let's have the cake first. Let's not assume that AI can just go in and magically bring things together. Build the cake, build your digital thread, build your data foundation, and then AI will do wonders. But that foundation needs to be in place. And when you have that foundation in place, now you can really go through multiple iterations of AI and really see value come out of it.

Chad Watt:

Christine, Ratish, thank you for being here today.

Christine Deliance:

Thank you, Chad.

Chad Watt:

This podcast is part of our collaboration with MIT Tech Review. Visit our content hub at technologyreview.com to learn more. Be sure to follow us wherever you get your podcasts and on YouTube. You can find more details in our show notes and transcripts at Infosys.com/IKI in our podcast sections. Thanks to our producers Christine Calhoun and Yulia DeBari. Dode Bigley is our production technician and I'm Chad Watt with the Infosys Knowledge Institute. Until next time, keep learning and keep sharing.