Hello, and welcome, everyone, to today's briefing on Moody's GenAI strategy. I'm Shivani Kak, Head of Investor Relations. Before we begin, I want to call your attention to the safe harbor language, which can be found in our most recent earnings release, dated April 25th, 2023. Today's remarks may contain forward-looking statements within the meaning of the Private Securities Litigation Reform Act of 1995. In accordance with the act, I also direct your attention to the management's discussion and analysis section, and the risk factors discussed in our annual report on Form 10-K for the year ended December 31st, 2022, and in other SEC filings made by the company, which are available on our website and on the SEC's website. These, together with the safe harbor statement, set forth important factors that could cause actual results to differ materially from those contained in any such forward-looking statements. With that, it gives me great pleasure to introduce our four speakers. We have Cat Tucker, Nick Reed, and Sergio Gago Huerta. If I may ask you each to briefly introduce yourselves, and I'll start with Cat, and then go Nick, to you, and then to Sergio. Cat, over to you. Thanks for having me. I am Cat Tucker. Of the many hats I wear, I am responsible for the implementation and integration of GenAI within Moody's Investors Service. Thanks, Cat. I'm Nick Reed, the chief product officer at Moody's Analytics. I've been working in fintech and financial services for 25 years, and I'm leading the approach to GenAI at Moody's, which makes sense to me that the CPO would be involved, given we think it's so inherent to our strategy and inherent to kind of the future offering, that we have, and I'm glad to be here. Thank you. Sergio? Thanks, Shivani. My name's Sergio. I'm really happy to be here today. I lead engineering, data science, and quantum computing at Moody's. Of course, this conversation is very dear to me and is part of my daily work I'm really honored to be in part in this conversation. Thank you, everyone, and let's jump straight in. I mean, we're here to talk. Mm About our strategic approach to generative AI, Nick, I was going to pose the first question to you. Can you tell us what Moody's has planned? Sure. Well, we believe Moody's is uniquely positioned when it comes to implementing GenAI across our entire business. We actually have decades of experience in analyzing risks using data science and machine learning and other techniques, and therefore, we think we have a clear window to be able to lead and deliver risk assessments in a way that help our customers anticipate risk, adapt to it, and thrive. In fact, we have some significant decision models in MA that have been using machine learning and natural language processing for many years. That's why we're focused on using the rapid advances in GenAI to help harness the power of our data, our analytics, our insights, and deliver them into new channels, and leveraging natural language. It's a powerful way to unlock our data and our expertise. Rather than limit access to the data and innovation to just a small group of experts, what we've done as a strategy is deploy it across our entire organization, as we want all of our employees to be able to embrace the technology. We've created 14,000 prompt engineers and 14,000 innovators and 14,000 testers, and we're encouraging our employees to be able to leverage what we're referring to internally as Moody's CoPilot, by empowering them to be able to experiment and to innovate inside a secure, internal generative AI environment. We're hoping that that will be able to foster a successful commercial implementation of GenAI as part of our journey. We've been working with traditional AI for a number of years now. How does GenAI change our innovation strategy going forward? Sure. As I mentioned, more than a decade we've been working generally in AI and machine learning and language, and also leveraging it to help us shape and develop some of our proprietary data assets, like Orbis. It means we have multiple products that are embedded already with AI and machine learning or algorithms. As an example, we have, in banking, early default warning alerts for more than 200 million customers as part of an EDF-X solution. In insurance, we can price contracts that are based on unbelievably complex catastrophe impact modeling calculations. In data solutions we offer insights into company structures and relationships based on sophisticated graph technology. As it relates to generative AI and the use of large language models, it feels like a kind of step change in the development of AI. We believe that given our kind of deep knowledge, and background in AI and machine learning, we can leverage the potential of GenAI and combine it with the power and the depth of our proprietary data assets. We actually believe we're unbelievably well placed to be able to take advantage of that step change. Actually, Nick, I've been in data for many, many years, and when Rob mentioned or referred to generative AI as a generational opportunity in terms of the impact on how we work, he cannot be more right. We see that in our own hands when we develop these models, but this is especially true for our customers and the way we deliver our products, our solutions. From my perspective as an engineer, the opportunity that we have in front of us when we are leveraging these large language models, and we use them to make easier for our customers to do what they have to do by using natural language, by using their voice or their text in the same way that they would speak with anybody else. This is what you can do today with the systems that you can use online. If you use these systems, there's a limitation. That is the grounding, the verification of the data, and that is our main asset here. When you're interacting with Moody's data, you can get this grounded, the truth, and only Moody's with our vast data sets can provide on the elements that you mentioned before, Nick. Whether it's properties, locations, companies, economies, news, transactional data. So that plethora of information is we can put together with large language models as are OpenAI and others, as a reasoning engine. I think this is the most important point, that we use these models as a way to be exponentially faster. A reasoning engine that gets its knowledge from our data sources, our data assets. At the end of the day, it's an extremely strong advantage here from what you were saying. We've been working with OpenAI, with these models for over a year, but for far longer in what we call today, classical machine learning. All the groundwork that is needed in order to start there, that's a head start for us. There's work in labeling, work in security, on entitlements. I mean, who can access this data, right? And who has the rights to access, and royalties, and so on. All that is done, and this allowed us to be extremely quick to leverage these 14,000 people and get them to play, and also with all the ideas that we have for the future. We have this reasoning engine. Yeah, and as you, as all of you know, we've been market evangelists in thinking about exponential risk. The idea that risk can no longer be thought or managed discreetly. GenAI is the perfect technology at the perfect time to bring risk concepts together using Moody's data, making correlations and inferences between risks to better inform our customers. To be clear, our intention is to leverage these technologies across our business. Yeah, that's absolutely right. This is a company wide initiative. We know that both MA and MIS have different priorities when it comes to the use of generative AI and how it applies to their business. We're excited about how we can harness the technology to make everyone deliver faster, to be able to more efficiently produce products, and to be able to think about ways in which we can include this into both existing and new channels. Partly that's why we made these generative AI tools available to all of our employees, because it gives our entire population an ability to have kind of direct and fast access to the tools themselves, and also to our vast datasets in a safe and kind of secure way. At MIS, our researchers and analysts can use it to assist in the process of drafting research reports, for example. Over time, we anticipate our analysts are going to be enabled to use GenAI as they would any other tool, just like Excel or other statistical platforms, to make that work easier and faster and more powerful, more efficient. Credit ratings are going to continue to be assigned by Moody's analysts through our rating committees, following our publicly available methodologies. Ultimately, GenAI is intended to enhance our ability to gather information, analyze it, and deliver it to our customers much faster. To be clear, we're optimistic about our ability to leverage GenAI while maintaining our analytical rigor and producing our credit analysis through processes with our analysts absolutely at the core. If I can add from the Moody's Analytics side, it's a lot about the way we access the data in novel ways in a much quicker ways. Almost like instead of going through 20 documents, and you get the answer that you're looking for, but not only in isolation, but getting it in a mix. Get both Moody's insights and knowledge from your own data, as in the customer data, in a private and secure way. Imagine basically getting both things together, a company data and Moody's different datasets from all the examples that we mentioned before. Of course, proprietary data is key to generative AI. That's table stakes for us. Making a very strong competitive position on helping the AI models themselves provide this verifiable and trusted information, is basically removing the what's known as hallucinations in the world of GPT models and large language models, by using our data to prevent that. Even then, we always plan to have, and this is very important, a human in the loop in the processes that we work with. These tools are to empower us, both ourselves, our people and our clients, and as a framework to evaluate the quality of the content that we produce, but with a human in the loop in there to verify and know those answers. Basically, and I'll finish with this. Generative AI will help us enable more and very much more accurate content much quicker, and make it easier to integrate with our own data and to reason with these data. This for our customers is a very strong competitive advantage. Cat, sorry, I'm gonna drill down just a little bit on MIS. Could you give some specific examples of how this could be used in MIS as a tool? Just to clarify, we're not using it as part of our credit ratings processes. MIS analysts and researchers can leverage our internal Moody's CoPilot that we referred to before, that's what we're referring to it internally, to use it in connection with research through internal tools. It gives them access to a plethora of LLMs, as well as Moody's public ratings and research reports that they currently use for their research. It's the same data that they have access to today, the history of ratings or industry research, they can now access it in a much faster and direct way. Moody's CoPilot can access and connect relevant data, news, and insights, which enrich the research MIS that we can produce in a secure environment. That's the most important part, is that we've created an environment where they can use it safely. For analysts, it acts as a Research Assistant. It's a CoPilot. That name was picked, not a pilot. Of course, the research we produce is still subject to the same checks and reviews as before. The authors are still responsible, so it's more of a resource acceleration and not a resource substitution. Thank you. I know we've had some questions from analysts and investors on this, I want to touch on the June 29th announcement regarding the partnership with Microsoft. Could we talk about what this means for our business? Sure. Well, to begin, we're pretty excited about this. As we mentioned in the press release last week, it's a strategic partnership that's gonna deliver on kind of next-generation data and analytics, and research, and collaboration, and risk solutions. Not just for those working in the financial services industry, but for anyone. Partnering with Microsoft has some pretty significant upside. We have a vast amount of curated proprietary data, analytics, and research, and we think by leveraging the power of the Azure OpenAI Service, we'll have the opportunity to unlock deeper, more integrated, more kind of on demand perspectives on risk. We believe the partnership is much more about being able to leverage those things and build together with Microsoft. We're both excited about the opportunities that will flow, that'll leverage Microsoft platforms and inject Moody's proprietary data into those platforms for customers of those ecosystems. We've been working alongside Microsoft engineers to start that development of these new products and to help us kind of envisage what we think that joint go to market will look like. I just want to add here that working with Microsoft has been an authentic pleasure. At the same time, these solutions that we're putting together are gonna be available independently. Microsoft ecosystems are fantastic, but there are a lot of exciting avenues to really explore going forward. Thank you. I know something that, analysts and investors have also been asking about is: What does Moody's have to gain, and what is Microsoft gaining in return? Sure. The partnership offers some pretty clear strategic benefit to both of us. We're both pretty excited about the value that we think we can deliver to our customers and to our employees. Moody's will benefit from the power and scale of Microsoft's Azure OpenAI Service. It'll allow us to kind of co-create new products and services for research and risk assessment. We think our validated and pre-indexed datasets and content will anchor Microsoft's large language models, injecting them with kind of more reliable outputs that we believe will result in fewer hallucinations, as Sergio mentioned. We're gonna be able to sell easily to customers that we have in common, and we're excited to bring a combination of their large language models and our content together to provide new tools for those customers, and brand-new customers for either of us. Of course, Microsoft will benefit directly from access to a broad range of Moody's solutions, including our Orbis database or applications in things like third party reference data management or counterparty risk assessment, or even in supply chain management. Taking a step back, how will GenAI differentiate Moody's from other market participants? Well, as you can imagine, as we've kind of been speaking to, is Moody's, we have a vast array of verifiable datasets and industry-leading content to leverage in LLMs. As the world of GenAI, you know, evolves, having quality, verifiable data and content will become increasingly important and valuable to our customers who need that verifiable data in their LLMs to deliver trusted outputs. These technologies can help us to enable us to more readily incorporate expertise from different areas of risk, and we'll be able to incorporate them into our ratings, research, and products. We believe we can improve the quality of the outputs of the LLMs to our customers because of our trusted and pre-indexed datasets. It goes beyond that, right? A lot of people stays at the surface on write an email or prepare a post for LinkedIn or other places. I think the magic here is when we mix the data together. Instead of thinking risk in silos, credit risk, supply chain risk, weather modeling, all the different things that we calculate today, whether with our classical models or with our data silos, now, with this reasoning engine, we put it together, and we move forward to a more integrated risk assessment with these proprietary data assets. The decision-ready models are allowing us to understand all the risk exposures from all the different angles. Since we… As I've mentioned before, since all the data labeling, all this modeling is already done, it's already there for the taking, for us, it's a very quick go-to-market to harvest these benefits. What are the products or the business areas where we'll see GenAI first? I'm gonna try to not spend two hours talking about this. Most of our business segments, in fact, most of our people are being impacted by generative AI today, one way or another, no matter their role. We're using these technologies, as Nick was saying before, 14,000 people as prompt engineers. Moody's CoPilot is the name of our platform, our internal tool that we use, and it gives these verified insights to our employees, regardless of what they do, whether they are preparing a marketing campaign, a sales pitch, or writing a piece or anything like that. We're encouraging our employees to experiment with this, and one example is that on Moody's Analytics, we have many channels where people support each other and trains each other on doing this. It's not only experimenting with how fast this industry evolves. There's every week new research that changes a little bit the landscape, but also how we come out with new use cases on a daily basis from the product managers, the product owners, even the engineers themselves. We do look forward to our customers seeing a big impact on research and insights for one, and then, of course, bringing the data where they can access the data from Moody's in a new way, whether it's on our current products or in feeds and other ways. Also mixing that data. By putting these things together across our products. I'm kind of echoing Rob Fauber, our CEO, what he said last week or a couple of weeks ago, the first potential generative AI product that we will launch externally will be a Research Assistant. This is something that I'm extremely excited about, but we'll be able to provide an update later in this year. In light of all the advancements in GenAI, how are we thinking about our talents and our assets? Yeah. As we've said, we're leveraging kind of traditional AI, and we've been doing that for years. We absolutely have assets, the assets that we need to succeed in-house. For example, we have many data scientists and engineers across the company with skills and the tools that they need, working on kind of prioritized use cases for our customers. We're really talking about empowering them, and empowering our entire employee population to be part of the GenAI journey. These 14,000 innovators and prompt engineers that we talk about will be able to help us innovate as they kind of actively explore the technology and create new ways to be able to deliver value. Really take it from a smaller group of specialized data scientists, and really be able to kind of exponentially grow that across the entire company. It is a very beautiful thing to see how we can permeate this knowledge and innovate together. When you had before a group of data scientists doing something in months' time, now they can experiment something very quickly in one day or in hours, and share that with their peers in different parts of the organization. Creating that sort of innovation flow has been fundamental for us. Now, one of the things that I wanna mention is capacity. Capacity at scale is one of the main limitations for all players in the market, everyone who's doing things with large language models. A shortage of GPUs is well known in the industry. That's actually one of the things that we made sure we secured with our partners at Microsoft and OpenAI, so we can actually deliver it based on these shortages. Our group, building on that, is not a centralized group that builds everything for everyone, but is more like an empowerment tool for the company that allows others to succeed. Seeing that happening in real life is extremely exciting. Cat, you've talked about an enablement tool, and Sergio referred to an empowerment tool. Does this mean, and it's a question we've gotten from analysts, is, you know, will GenAI make either the analysts or our ratings obsolete? Absolutely not. We can't say that more definitively. We see this technology as an enabler for our employees, including our analysts, giving them so much more capacity to focus on higher impact work that requires creativity, expertise, and high level reasoning, which is key. We also believe this technology has the potential to enhance the ability of our analysts to assess credit risk, making our ratings even more valuable in the market than they are today. This is really a fantastic opportunity for us. Thank you. Leading on from that, you know, what are the revenue opportunities and the opportunities for efficiencies with GenAI? Sure. It's a little too early to quantify and give a specific number to, but we clearly believe that there's potential for revenue uplift here. The uplift opportunity is probably going to not just come from enhancing our existing products, but we also think there's an ability for early opportunities to be created for new channels and new ways for our existing customers to access their existing products. Also some pretty exciting ways to use those new channels and tools to gain new customers, including inside the Microsoft ecosystem of Teams and Office 365. We also think that there may be new revenue opportunities by reducing the barriers between our products and allowing our customers to use natural language to ask for what they want. This, we think, is gonna create both cross-sell and upsell opportunities and also expand the type of customers who will be able to use our product. By using natural language to kind of democratize access to the various components and products that we have is part of our strategy. The cost side of the equation are there any savings to talk about, and when will we start to see them? Sure. As you invest in new capabilities, naturally, there's gonna be a kind of cost element associated with this. We strongly believe that when applied strategically, GenAI has the ability to boost productivity, improve workflows, and improve the overall operational efficiency, not just for Moody's, but also for our customers. We have done a lot of experimentation and research in this, in this side. What we consider is that on our current processes and products, some of the workflows that we have been exploring can give us a fraction of the development cost or time that we had before. Some of use cases that we have researched internally can give us up to a 40% time savings or efficiency gains on doing that. Also recently, I'm really happy to share, we've run a large experiment with a fraction of our software engineers on a coding copilot. That provided even better numbers than that we'll be able to share, we will be able to share very soon. Basically, we can increase our throughput, we can increase our efficiency, our accuracy, and help our employees focus on those value-added tasks instead of doing the repetitive ones. Those are the things that we engineers don't like to do. How this translates into specific cost savings, as Nick would say, it probably is a bit too early to determine. It's, I would say, just one of the aspects of everything. This is like, We're seeing how this is improving our products and our own efficiency, even for myself. We'll be able to measure that very soon. I can tell just from your, just the excitement and passion that you all have for GenAI, that there is a lot that you're excited about. I want to just turn, Cat, first to you to ask what excites you the most about the GenAI opportunity for Moody's? GenAI puts the power of a large language model, which is designed to create and harness infinite data connections, in the hands of an analytic team that is best-in-class with connecting data to develop insights and assess risk. I really see it as an analytic superpower. Sergio? I'm really happy that you asked me this question, Shivani. I think I've worked now a bit more than a couple of decades in data and AI, and a few years ago, I decided to switch my career into quantum computing. I was lucky enough to build a quantum computing team at Moody's, and we've been working on that. That's kind of a horizon three. It's a bit later in time for this opportunity. AI, in the last year and a half or so, has pulled me back into this, because really, this is a once in a lifetime opportunity. I can talk from an engineer, with an engineer hat here. Being at Moody's and doing this is like being at the right place at the right time, for all the reasons that we've mentioned today and I'm really, really honored to be able to do what I am doing. That's great. Nick, just turning to you finally, just to recap for you know, what are the key takeaways that you think the audience should take with them from today's discussion? Okay, well, I guess picking up from Sergio, firstly, we've kind of talked about the strength in operating in an environment where we're already focused on exponential risk. We think this fits right into our wheelhouse. We believe we're uniquely positioned, kind of courtesy of the integrated risk assessment strategy that we have and the validated and proprietary data assets that we have, to utilize GenAI across multiple risk domains. Secondly, our vast and curated data assets and content are already ready to go. They're ready to be leveraged in conjunction with large language models. We think that gives us a head start. Because our content has been verified and validated, we think we're gonna be able to ground the output and reduce the likelihood of hallucinations when our customers are using these kinds of tools. Thirdly, our strategic partnership with Microsoft enables us to co-create new products in a secure and reliable environment, and open up new channels, and deliver those products and solutions, including in Teams and Office 365. I guess finally, underpinning all of this is our approach to how we're implementing GenAI across the firm. As you've heard over and over, we've been working in more traditional AI and machine learning for decades, the GenAI opportunity has meant that we've kind of mobilized around this. We've kind of impressed on the whole firm, really firm wide, to embrace this technology, and we've allowed them to be able to really lean into this. We believe that that agile mindset and that openness to innovate is gonna be really core to our success. Well, thank you. I think that's a great way to end the call. Thank you everyone for your time and participation. Thank you to our audience for joining us today. All the best. Thank you.