Video: From Information to Intelligence: New Disclosure Signals and How Finance & Legal Leaders Are Putting AI to Work | Duration: 3764s | Summary: From Information to Intelligence: New Disclosure Signals and How Finance & Legal Leaders Are Putting AI to Work | Chapters: Welcome and Introduction (37.66s), Opening Remarks (115.075s), Measuring AI ROI (1229.48s), Human Judgment Redefined (1618.065s), AI Process Transformation (1758.195s), Cultural Transformation (2089.065s), Leadership by Example (2462.56s), Hands-On AI Learning (2542.275s), Agent Orchestration Strategy (2738.715s), Behind the Research (2941.67s), In-House AI Efficiency (3172.13s), Board AI Governance (3430.205s), Domain-Specific Tools (3568.455s), Closing Remarks (3734.965s)
Transcript for "From Information to Intelligence: New Disclosure Signals and How Finance & Legal Leaders Are Putting AI to Work":
Okay. Good morning, good afternoon, good evening, depending on where you're you're calling in from. We're excited to to have you here today for a Nasdaq webinar, from information to intelligence, new disclosure signals, and how finance and legal leaders are putting AI to work. One housekeeping item before we get started, a disclaimer, to, to let everybody know that, the the folks here presenting today are are not attorneys. We actually maybe have one attorney and Mark Casper, but nonetheless, this is not legal advice, and to consult your own attorney for any information around disclosure or tax or professional services. Second, some of the insights are, delivered through a solution called Nasdaq lens. Nasdaq lens is an AI powered solution, and it's used to generate some of the content you'll see here today. AI can make mistakes and so verify all outputs before using the insights. I'm gonna pause here for about ten seconds and let the the audience, read the disclaimer. Okay. Well, let's, let's kick into the actual agenda. And with that, I'm gonna bring on today's moderator, Christina, and she'll take us away. Christina, welcome. Thank you, Mike, and welcome, everyone. My name is Christina Ayanian. I'm Nasdaq's market site reporter, and it's a pleasure to be with you all today to moderate today's insightful conversation. So to set the stage, we all know that AI is moving fast. Some argue that it's moving too fast. Others say it's not moving fast enough. But the common theme is that it's gone from something that companies were experimenting with to something that's now integrated into the everyday workflow, especially in public companies. And we're seeing it show up in some very interesting places, in disclosure statements companies share with investors, in conversations that are happening in the boardroom, and in the overall priorities of finance and legal teams. I always say, in my opinion, it's turned every leader in an organization into a partial CTO. And today, we are going to explore both sides of that shift. So we'll first kick things off at the new findings from the disclosure report, which is Nasdaq's research examining the 2026 filing season at scale. We'll dig into how risk factors are evolving, what companies are actually saying about AI in their MDNA, how boards are approaching AI oversight and director education, where AI is starting to appear in executive compensation, and lastly, some recent, survey data on how finance and legal professionals are using AI themselves. Mike Siller from Nasdaq is going to walk us through all of those findings. And from there, we'll move from what companies are saying publicly to what's actually happening inside their organizations. And I'm very pleased to be joined by Sarah Youngwood, chief financial officer at Nasdaq, and Mark Casper, chief legal officer at Marvell. We'll speak with Sarah and Mark about how their finance and legal teams are approaching AI, where they're seeing real tangible value, what they've learned throughout this journey, and how they think AI is going to reshape their team's workflow in the years ahead. A little later, we'll go behind the research for a deeper look at how Nasdaq's new purpose built AI research agent is analyzing public company information at scale that honestly would have been very, very tough to pull off the traditional way. And then we will bring everyone back together to answer your questions, and we encourage everyone to send their questions, through the q and a function throughout the session. So we have a lot of great material to get through. Let's start diving in. Mike, we'll start with you. Let's start with what the research is actually telling us and what Nasdaq is seeing across the market. What stood out to you and your team as you analyze the 2026 filing season? Yeah. Thanks, Christina, and, super excited to to take the the folks through the research. Before we do that, I want to spend just a minute on, how we created this research, because I think the methodology is as, as important as, as the story. So the disclosure report, that, powered... Was powered by the disclosure research agent, which is within Nasdaq lens, a product that that Nasdaq has. And so the agent was designed to read, understand public disclosures at scale, not simply search for keywords, but analyze the context and meaning within filings. For this report, we analyzed 42,000 company filings across 4,500 companies, including annual reports, quarterly filings, proxy statements. We then use the agent to identify patterns across that content, themes that were becoming more prevalent, changes from one filing to the next, differences across companies, industries, and emerging areas of risk. Importantly, this research was simply... Was not simply generated by by AI and and published. The human layer is important throughout the process. And so NASDAQ domain experts reviewed the findings, interpreted the results, and refined the analysis. And we complemented that filing analysis with a 150, legal and finance professionals that we surveyed to understand how AI is impacting their day to day work. So you really see a combination of things here throughout the research. One is Nasdaq's unique data, two, AI enabled research at scale, and expert interpretation. Then we put that all together in a in a nice report that you can all consume that we'll share post post webinar. So let's dive in. I think the big headline that that we identified from the 2026 filing season, so think about what what transpired between, call it, January through through June, was was AI is crossed from being an emerging risk topic to a material business topic. Right? And so let's start with risk factors and what we saw from from the twenty twenty five, twenty twenty six season. We saw that four eighty eight companies added new AI risk factors to their 10 ks. And so that is a material jump that, that we had not seen in previous sessions in previous, previous years. And I think the big thing was it wasn't just defined down to tech and software related companies. It became broad based beyond that, that that technology based community. So you saw manufacturing, you saw health care, you saw all... A large swath of different sectors start to add these related, risk risk factors to their to their disclosures. Wanted to share a bonus insight, that you won't find in the report is what we're seeing already in the twenty twenty six ten k season. So the early reporters, everybody likes to look at the early reporters and learn from them to inform what next year looks like for them. And so we've already seen 73 companies add new risk factors in their twenty twenty six ten k so far. The reporting season obviously just started for a lot of companies. And so we continue to see the same trend that we saw last year in terms of risk factors being added to to, to, the 10 k's. AI risk factors being added to 10 ks. So I think, what I would say is that's just half of the story and probably the more interesting part of the story is what we're seeing in MD and A sections. And so, increasingly m and d, AI is showing up in the MD and A section. And, you know, I think what it's telling us is the, the conversation is moving beyond risk to, to opportunity or business strategy. And so when we looked at the MD and A section and, and the AI mentioned within those 32% appeared in the executive overview, which we thought was an interesting trend. That was by far the most of all the different categories within MD and A. And I think this has been significant because the executive overview section is typically where management talks about, strategy growth, investment and competitive differentiation. So the fact that AI is starting to show up there is is really important. And so today, we saw it very popular in the software and IT services and semiconductor space, as you could imagine. But But, again, similar to what we saw in risk factors, it's actually starting to pop up in what I call more traditional sectors like manufacturing and financial services was definitely, an area where we started to see AB show up more consistently within within MD and A sections and particularly the, the executive overview section. So, you know, what was once ago... What's... Was one year ago or two year ago, disclosure question is do we need an AI risk factor has really changed to is AI part of our story and how consistent are we being across the filing from risk factors all the way through the MD and A section. And so if MD... If AI is a is a strategic opportunity in MD and A, then what are the associated risks? Right? How is a company governing that? What are the investments that are being made and are those different pieces consistent throughout the disclosure story? So as we move on to, that same part of the story, which is okay, if there's risks and there's opportunity, well, then how are boards governing board... Or governing AI oversight or governing AI at the at the board level? I think there's some really interesting trends that that we were able to discover in the disclosure research agent was able to pick through for us. So we've looked at a 175 mega cap companies, in their twenty twenty five, twenty twenty six proxies. And, what we identified was 50% of those companies now have AI board oversight, at at that board level. So AI oversight at the board level. So 88 of the one seventy five have that. That's up from 33% in in 2020, 2025. And so 58 of the one seventy five, so 30% net added an AI oversight at the board level. So 52% growth. And so what you're seeing is it's really being embedded into existing structures. You're very seldom seeing an AI, AI committee being being generated new. It's typically gonna be the full board oversees AI or it's embedded into the audit tech technology risk or cybersecurity committees. And so very infrequently is it gonna be an AI specific committee. Another thing that we identified was, executive compensation. And so only four companies, 4% of of the companies in that larger cohort of one seventy five, have, AI at the... At a... As a quantitative metric in executive comps. Still very rare. You see... You see... You do see it show up slightly in, in more of the, the more general qualitative, executive comp metrics, but not not so much at the quantitative level, yet. I think the last thing I would share just from board oversight is director education. Only 17% of the companies that, looked at with the one seventy five, discussed the... Disclosed formal AI, training at the board level. And so maybe it's happening, but not being disclosed, but we felt that was an interesting finding that the research agent was able to pick out was only 17%, are are communicating that they're they're educating their directors on an AI basis. So we'll move to the last, interesting insight that that we've gathered. And this was not from the disclosure research agent that you can't find this in in disclosures, but it is part of a survey that we performed across senior legal and finance leaders this summer. And we asked the question, where is AI delivering the most value? And so what we found and what the community told us was research is the is the number one topic. And so 62% said speeding up research and analysis was the the most important or most common, benefit that they're seeing today. As you move down here, you see that 15% actually said they had not adopted AI yet, which was a surprise to us. I think if you probably ran the survey again, three more... Three months later, you might see a different number. You might see the 15% go down pretty pretty dramatically. But improving disclosure was a general topic. Drafting, was another topic that came up on a consistent basis. So, you know, I'd say it starts with starts with research. It's transitioned into... It'll transition into drafting, but, you know, research is the wedge to to really get started. So with that said, we'll, we'll bring back, we'll bring back, Christina here to to kick us off for the executive conversation with Sarah Younglood and and Mark Casper. Really excited for that. Hopefully, the disclosure research insights brought a little bit of perspective around what we're seeing in disclosures, what's important, and ultimately how the finance and legal professionals are using AI today in their work. I'm excited to to hear from the leaders here from from Nasdaq and Maribel to understand, what what they're hearing from and seeing from their own organizations. So, Group, welcome, and, I'll leave it to you. Thank you, Mike. You know, the numbers speak for themselves. We've come a long way, but we still have a long way that's what companies are seeing publicly. Now let's dive into what's happening inside the actual organizations themselves. So we're seeing AI show up everywhere in risk factors, MDNA, board oversight, executive conversation... Compensation. Stepping back for a second, Sarah, what does that tell you about where AI sits on the executive agenda now? Absolutely. First, Deepers, thanks for having me, helping me. and I'll meet all of you. What I believe is this tells you is very consistent with my experience, which is that it's extremely high on the executive agenda. We are seeing, AI as a fantastic avenue of growth. And, so, of course, executives, at all companies are gonna care about increasing their opportunity set, that is, new products, new new servicely addressable markets, new total addressable market, and also changing the way one operates, operating with more efficiency and being able to handle more. So whether you're looking at it, we call it in the products, which is really continuing to improve your products so that you, maintain, your role as a transformation partner to your clients, or whether it is on the business, which is what, Mark and I and others are doing, which is transforming our functions. AI is absolutely everywhere, and I will say, I hope we were in the 17%. Our our, board is getting tremendous education on, Gen AI, and it's, many, many conversations. That's it. It is not a training. It is really part of the discussion. And it's safe to say it's it's here to stay. Mark, how does that look from where you sit? Yeah. Thanks, Christina. It's a pleasure to be here. So entirely consistent with Sarah. I would say it is top of mind for both our board of directors and our executive team. And it really starts, I think, in r and d, and that's the real game changer for companies like Marbell. Because if we can cut down the development or cycle time, say, from two years to one year, I mean, that has a dramatic impact on our business and gives us a competitive advantage right there. But it doesn't stop just with that side of the house and also for the supporting organizations is big. And so it is a topic that, legal, accounting, compliance, trade compliance. We are expected to talk about AI initiatives at every board or committee meeting, and it and it it starts with, here's how we're using it and why, but then where it's going is metrics. Do you have... Can you measure how it's making you better? Can you measure the efficiencies? Is there an impact on outside legal spend? Is there an impact on headcount? That's what's coming next. So adoption is there. Now I wanna see the metrics behind it. Sarah, you called AI an embedded capability, not just a side project. And what I think is really interesting, on top of that, you've put your entire finance team through an AI proficiency program with belt levels. What changes when AI fluency becomes an expectation and not just an option? Absolutely. I think that when you, think about, GenAI, we've talked about the fact that it's an extraordinary technology. It's here to stay. It's everywhere from the boardroom to the management committee to the products, to the product development, to on the business in all of the parts of the business. So sometimes you need to push people to do what is actually really good for them. So we made it mandatory, to actually, change. And and the change is a change in culture that we're trying to affect. This belt program has three dimensions. We said, okay. You need to train yourself, and we're gonna tell you what training means. You need to engage. I use the platform, and I will tell you how much engagement is needed. And you need to deliver. You need to actually deliver agents that create value. And so for a white belt, which is our first level of belt, and I'm glad to say that we are at 77% of my organization that has, a white belt. So we're doing really well in terms of progress. We actually are, requesting a small amount of training, a regular but not, permanent, level of engagement in the platform, and not yet, not... Nothing yet in terms of, like, quote value added. You might use it for personal productivity, but you don't have, like, all of those great agents that have been vetted by your manager as actually having a value add to the company. And then I'm not gonna go through all of them, but by the time you become a black belt, which is meant to be a small percentage of the organization, you are not, taking the training but teaching the training. You are, living through, the GenAI platform, and you are adding tons of value, and that has been recognized, by your leadership team as such. So, we've got several levels of BILT, and what's great about it is that it's very tangible. Especially in the finance function, it's very good to be able to measure, and people are used, to things. And what's good about it is that it's multidimensional, and it enables us, to make sure that we embed the behavior within the teams and that we give the opportunity to the teams to actually change themselves, so that they become super relevant tomorrow. And it's fabulous that we have one of our, top persons, who is, a black belt and who is a tax person. It's incredible, to see, how this technology requires curiosity, requires, of course, a very rigorous mind, but does not require a PhD in computer science. I would definitely love to do that training. So, it's interesting because you measure AI like a return on invested capital. So engagement first, then velocity, and then financial impact. Where are you and your team in that journey today? And, furthermore, what do you tell a c... A CFO who's asked to fund AI before that stage three, that financial impact section is even provable? Sure. So when we think about return on invested capital, first of all, this is something that we use not just for GenAI, but that we use for all of our decisions. We first need to have a strategy to be able to have a framework, for the decisions. And so, our strategy very clearly is to embrace Gen AI, but not just that. Gen AI for us is gonna meet digital assets. It's gonna meet continuous markets. And in the construct of those three trends reinforcing each other, we need to know what products are gonna be needed, what type of value add we can have for our clients, how we can operate and add, to our role in the markets of tomorrow. And so we make decisions, that are not just about GenAI, but about how to invest. For GenAI, you said it very well. And... But we call it inputs and outputs. We start with the inputs, and you can think of that as foundation. It is a foundational aspects to be trained and to engage with the platform that it was nothing. But it's the first layer. It is foundational to make sure that you have the right trust embedded in your platform. But once you've done all of those things, you have a chance to deliver value for shareholders literally today. We are selling products, for example, financial client management agents to our clients today. So I don't need to have, like, a strategy that is ten years, forward to know that if I sell it for this price with that many that many clients and I have that much of a pipeline, that is a good ROIC. But then we... That's horizon one for us. If you look at r and d or what is much more forward thinking, we call them horizon three. It could be, like, five years plus. Incoming. That's where embedding it with all of the other trends in the business is really important so that you are not just pushing something that makes sense from a Gen AI point of view, but actually corresponds to what we believe is a lasting need in the future landscape. Mark, how does that look from a legal perspective? What does AI unlock for you? Is it saving time, money, or is it something completely different? Yeah. Actually, it's a little bit different for us. It's... I've noticed the quality of the delivery of the output of work is is improving for our team. And it starts even with communications, internal communications. They're much more clear and concise. And typically, if you think of Marvell, it's an engineering company. Okay? Lawyers like words. That's how we communicate in a lot of words. Engineers don't like words. They don't understand them. They like tables. They like graphs. They like charts. So even that output of taking words and be able to put it in a language that our clients, our engineers, understand. Right there is a huge game changer for the for the legal department. We're also seeing, the ability to do, to to capture many efficiencies in our work. So from our corporate practice, say for example, if we we, want to adopt... Make some changes to our insider trading policy, we have our peer group already loaded up into our AI platform, and we're able to quickly do that analysis to match up where our peers to the extent they make it public. And so that's an... That's a a project we can do internally very quickly and then take the output into a a format that everybody can understand. And then I think, the same thing you'll see in our risk factors where we have developed some new risk factors, and we're able to use a platform that will look at other companies and, hey. We wanna model a risk factor after this, but modify it. I think we're seeing the output, of that, not only the quality, but the efficiencies as well. We we touched on AI moving from experimentation mode to something that has real tangible value. Where have you seen on your team that AI has moved from something that your team was just testing to something that's actually changing how work is getting done? Sure. I'll give you a couple examples. So we have a sustainability team, a really important topic for our hyperscale customers, and we will get a number of sustainability questionnaires that they want answered as part of the RFP process of party... Of of having us do a project, and it's ongoing. And, we used to have to do those manually. And now we've trained an agent to be able to, take the questionnaires and provide baseline level answers. So it went from something... I think it was about two and a half hours of time from the team now down to fifteen to twenty minutes. That's one example. Another is with our commercial team with contracts. We have, a, a channel where our clients can... It's a Slack channel, ask questions. Oh, I, you know, I need a three way n d... NDA. How do I use that? Or do I need a contract for this purpose? And, really, we have a bot answering those questions and with the ability for human oversight to jump in at the right time if if if it's if it's appropriate. And then last, I I use it. We developed a, sort of a platform with board and committee agendas. So essentially, there's standing agenda items that we have at every board meeting, every quarterly board meeting, and we trained it to look at the last four years. And so now we'll have an agenda get sent to me, a draft agenda, forty five days in advance of the board meeting that captures the the normal cadence of items that we would have at that particular board meeting. And then if I remember to load up my notes from the prior board meeting, it'll it'll have the new sort of subjects that the board, you know, wants to be, updated on at that particular meeting. And, so those are showing me just the ways that we're we're using it. You mentioned human oversight. There's a lot of of debate right now as to where and when should humans come into play. So as AI takes on more research and first drafts, Sarah, where does human judgment actually matter more and not less? I think that human judgment is very important that define judgment. And I'll give a few examples. In a forecast, the judgment comes in making sure that we have the view from the business as to what is happening. It's making sure that to extend that we have a change in strategy. We reflect that in forecast. But if judgment is not to make up an assumption, as a junior finance person when actually there are series of numbers, that could actually be a much better assumption than the one you can make. And people might call that a judgment, but it is not a judgment. And so I tremendously believe, that you're gonna continue to want human judgment, but that, like Mark, there is the automated judgment, which actually is more reliable than the bespoke judgment, that is actually not a judgment, but an inference that is based on experience. And we can give that context and that experience to our GenAI platforms so that they have the white contacts, the white, quote, judgment. And we live to the human judgment what is truly, what, years of, experience, business acumen, change in direction, things that cannot be picked up, by the agent need to, inform. And so I believe a lot in, the inputs are gonna define your outputs. And if you give your platform the context, then it can have a first layer of what used to be human judgment. Mark, did we lose you? We'll we'll come back to Mark. Sarah, the the research report was produced by an AI agent re... An AI research agent reading more than 42,000 filings across 4,500 companies with Nasdaq's experts validating those findings. So what is something that your team can do today because of that scale that would have been impractical a few years ago? Yeah. I think it's fabulous. I mean, what's very exciting, and my team is very excited, is, that you can think of GenAI as a lens, and I've talked about that. At the beginning, you have, like, the first level, which is general, like, personal productivity. Fine. And then you start thinking, oh, I can do this little automation. And then you go, wow, this thing can read faster than I do. It's actually remarkable how fast this thing reads. And and so you can do better research. And that's the beginning layer. And then by the time you have gotten enough experience with it, you end up with the, oh my god. I was thinking of this process as automating the current process from a to c. But in fact, that process go from a to e, and I should start with a step d to e before I do a to b. And and at that point, you're re engineering a process. You are making the process smarter, and you are reserving your human judgment to where it belongs, which makes the work actually much more interesting. And by the way, coding agents is very fun too, and our team is loving doing that. Do you think the biggest change is speed, scale, the quality of the analysis, or is it the ability to ask a bigger question than we would have thought of in the first place? Or all the of it. It's all of the above. I I think that you definitely have an element of speed. You have an element of, precision. I I do believe in what what Mark is seeing. You are... It's as if you had, like, this great analyst who is reviewing everything, and now you can go to... From a quality control, that is sporadic to a quality control that is much better than that. And that can be a 100%. We're saying that in audit, for example. To, sometimes... Actually, I love to to reason, with my GenAI. And I also, sometimes, like, let's say that I go to a play... To a meeting or to a round table and I take notes. I take quick notes. And then I verify the facts that I got before sending them out, which is amazing. So I take steps I wouldn't have taken, because it's easy to do. And then on the the big picture, you can't just ask GenAI what is my next step in my strategy, but you can actually, find some insightful comments if you say, I'm trying to do just this this and that. If you prompt well, you can get some fabulous insights because it has a library of the entire world of experience. Mark, what has to be true before you let AI into a high stakes legal environment? Yeah. Good question. Well, and and let me let me tie this back to, the human judgment piece and experience. So for example, company... We we... All companies get litigation, and you get to a point where do you settle or do you go all the way through trial? And we have a very disciplined process that we look at different points during the litigation and we actually will say, okay, we can resolve the the litigation at this point through a settlement for this amount, or we can go all the way through a trial or an arbitration hearing and here are the here are the different outcomes and potential, scenarios for liability. And, we've actually used that in an AI platform to assist, hey, with what's the right decision, settle or take this to trial. Now the human element, you need to be involved because there may be implications outside of that decision. Like, oh, it could be a very public process if you go to a trial. And, oh, your customers would prefer that you resolve this confidentially, or you might set a precedent one way or the other. And so that's where I think the human element comes in, but these platforms are enormously effective just doing the analysis I went through in a in a litigation. And they actually improve. The more you use it, the more the results, get better. So does does source linked evidence or, you know, being able to click back to the original filing language change your comfort level? Absolutely. And really for the team, that was the game changer for the for the team that are using this. The fact that they can click on that source to confirm, the the the the results that they're getting, that really instills confidence in the team to to use it more and more, to have that original source, whether it's a case, whether it's a site to another company's policy. But that... That's really... That was that was a significant game changer for those that were hesitant on our team. Sarah, is the bar different in finance? I would say, I'm loving what I'm hearing from Mark. It's, like, so similar, and it's also the same notion of we are not, like, reckless at all. We think that, our organizations have tremendous trust embedded in them and that we therefore need, to uphold that brand. And so we need to know what the source is. We need to know, how the human judgment agrees or disagrees. By the way, I have sometimes argued with my GenAI platform, and, the argument could go like, but have you considered this or have you considered that? And, when you consider more, the the platform might say, okay. In that case, I would, change my mind. It's actually really interesting because, you can't take all of it as ghost spell, whether it's finance, legal, or, even code. You want to make sure that, you debate as we all do in meetings. And you can debate now with one more thing, which is a series of agents. And I think that's a very important, routine. We also have, like, some agents that are specialists of debates. Like, literally, you tell them your job is to take the counterpart, and you will be the devil's advocate on this and this and that. And if you create your devil's advocate agent, you're gonna learn something. And so I think that whether it's finance or not, it depends what we are trying to do. But there is tremendous value that is being done by the agents, but not to the point where we believe, that, we should retire human judgment. That... That's a really interesting point. And going off of that, what what has been harder in practice? Is it getting the technology to work, or is it changing how people work? To Mark or to Sarah? Sarah. We'll start we'll start with Sarah, then we'll go to Mark. Excellent. I think that, so it took us a while actually to get, about a month ago, we were at 55%, on my team on, on the white belts. And the white belts is a very small requirements. And I was getting worried that after six months of the program, we had only gotten to half the people or just about that. And somehow in the last months, we got, like, people embracing it and, and and we are starting to see the snowball. And I think that what's extraordinary about this technology is that every month is an exponential, in terms of the progress that you're seeing, both in terms of the capacity of the models, the capacity of, the, different platforms that are leveraging, the models, but also in terms of your team becoming more comfortable. The unknown, becomes the known. And then, it's a very, engaging, tool because it uses human language. And so, it is, easy to adapt in some ways. So I think that the... It it is still, I think, the most important part and the most difficult part is to get culture right, which is why we're so focused on those belts and, we discuss it tremendously. The management committee and that the board, it's changing the way you work, and it's stopping to debate whether the technology works. The technology works. That is not the question. The question is, are you asking the right question? Are you giving the right context? Are you debating appropriately with the technology? Are you feeding it, information, in a trusted way? Because, obviously, it's all about your data. And so if you cannot, for security reasons, give the information to the platform, then the platform will not perform. But it's not because the language model would not perform. It's because you don't have it. And, for example, what you guys do, as part of, the product that you just demoed is you categorize each sentence so that it can be read well by the language model. That adds tremendous value because then you get a much better output based on the inputs, that you get. And so if our data internally is very well cataloged, and if we use tools, that are also very well cataloged, we are able to get from the same exact, model a much, much, much better result. And so for me, it's about changing the culture, changing the way people work, giving people the opportunity to do that, but also, having a cloud and data strategy. We're in some cases, which you can, acquire, but in some cases, in most cases for your internal data, it's your cloud data strategy that enables you to feed correct context to those very intelligent models. not what it is. Very consistent. It's it's it's the people that was the most difficult to change and change in mindset. And so I think Sarah said, we have a term where we call it embracing AI throughout the organization. And that, needs to be... We communicated that early and often to all employees. It's something I talked about with my team that AI is your companion, your partner to make you better. If you ever feel like it's going to replace your job, I guarantee you I'll get another job for you, so don't don't feel that way. Lawyers tend to have an, you know, just... Or most lawyers that, okay. I've... I'm highly educated. I went to law school. I've worked on thousand transactions. There's no way an AI platform is going to be as good as me or supplement my expertise. There's there's some of built in bias a lot in the system there. And so that was really difficult at first to overcome, but, we got everybody on board. And I think I have to lead by example. I think that's a job of all our our leaders. And so I'm in... I'm talking about it every day. I'm in the training sessions. I'm giving examples of how we should deploy it, and it's consistent. And, so now we've got the the full team on board. And, but that was the biggest challenge. Technology's Can. I, Yeah. Yeah? I just want to jump in to violently agree with what you just said. What you said about your leadership and role model, I think, is the key. We are finding that it... The management committee needs to model it, not just tell people it's important, but actually do it. Take the training, know how to use it, get their belts for us. And so I just wanted to reiterate what you just said, Mark. Sorry. I interrupted with that. Thank you. Well, Sarah, you're you're actually going through the AI belt program yourself. So what did doing that work personally teach you that a briefing wouldn't necessarily have? So I find it, difficult to learn without doing. And, and so, I think that this technology is one, again, that lens that is very iterative. And so at the beginning, you get through the first levels of, like, oh god, it's a technology that does not work. It's not that it doesn't work. It's that it... Your iPad is not configured correctly. The pop ups are not, enabled, blah blah blah. And I'm doing it on an iPad, which does work, by the way. And and so by the time you have, gone through the first things, you first think it doesn't work, but it's not because it doesn't work. It's because your configuration is not there. Then your coach basically fixes that for you. Now you're, like, in session three and, everything works. Now he goes, like, go do something, but you don't know which button to press. And so by actually having a coach and by seeing that it's here and there, or by learning some of the ways of prompting, like, how would I have known that you could have, like, a debating agent? Well, I was taught. And so... And then you start... Then you you start understanding. Then you go like, okay. I'm always talking about the same people. I want a skill and contacts. How do I write a skill? That sounds like... I don't know to find out it's explore blah blah blah blah blah. Like, there is a way to find the things. And then, I give the feedback that even with a coach, it was really hard to write skills in our platform. Next thing you know, they gave me a, here is your help. This is the agent that helps to create the skills. So I used that yesterday in our new training session, and I got it done. And and so it's really important because, first of all, I consider that I am smart enough, to be able to use something if I am taught. And when I can't figure it out, I say I refuse to tell my team that it's doable. And then I have the platform fix it. So so, a, us senior people, presumably, we were not chosen for being dumb. And so if we can't learn it, it's not learnable maybe. And, certainly at scale. And so we're not asking me to write to the code, but we're asking me to use the platform. And if I can use the platform, it it is usable. And then I can declare it's usable because I am past the white belt, and I've moved on to other belts behind it. It's hard for people to argue that I have more time than they do. And I'm sure, Mark, you feel the same. I do. So so if we have this conversation again two years from now, what work that is being done manually today would surprise you if it's still manual two years, later? Mark, we'll start with you. Yeah. Well, first, any manual process, anything that a team does, that that goes away. Everything has to be automated. That that will be, table stakes. You know, for example, I think all of us remember we used to set up our own spreadsheets and our models go into Excel. Like, no one does that anymore. Like, you can just go into your platform, it'll do... Tell you what you want to set up. And and, so I just think with everything that an internal legal department does, you know, from expense reporting, that's another example. All that will be will be automated. Zero. I agree. I think that there is no manual work that should not be automated because, the agent takes you from a to b extremely well. That doesn't mean that the entire judgment, is not needed at some point, but the agent will have embedded in the workflow to ask the humans the questions whether humans need to interact. They will triage, for example, our sanctions agent, looks at all the, sanctions that are prepared for clients and says, 80%, boom, they're prepared. You don't need to touch them. On those 20%, I need your take on this and that. And so agents today are already capable of doing things like that and triaging, so that when the human comes in, they come in for the right thing. I also think that what's interesting is that, as leaders, we need to make sure that people are not just building their patchwork of different little things. And so we need to make sure that there is strategy and that agents are orchestrated. And so, for example, we mapped all of our core processes in finance to say, it's not just that we want to reduce the manual inputs, but we want to automate a to z everything that's possible with the right human inputs wherever it's needed. Okay. For this process one, it's gonna take 17 agents. And so we build agent one, agent two. At some point, we are at agent 15. But when people build, they are building as part of the architecture that we have given. And so Gen AI, like anything, needs to be architected so that you get the most value of it. So the small automation, all of it. But even the very large automation, even if it requires human input, all of that can be agentically orchestrated. Yeah. And and maybe I'll add one more point from, from the internal legal department perspective. The whole, patent prosecution process from invention disclosures to patent committee reviews to drafting a patent. I think in today's environment, there's a lot of manual inputs into that process, and that's where we're starting to develop agents. And and that process is is time and intensive and also from a cost perspective, it's it's it's significant. So that's one specific example I think that is not been tapped by many, but it it will be. Yeah. Well, thank you both so much. This was super insightful. I'm looking forward to speaking with you guys again on the q and a. I don't think there could ever be enough time allotted to speak on this subject, but we just talked about, the research, what the research is showing, and how AI is changing work inside finance and legal teams. So now let's see how AI helped generate the disclosure insights from the disclosure report, in a video called behind the research. Earlier, we shared findings from the disclosure report, our analysis of how public company disclosure, governance, and AI are evolving. Now we wanna take you behind that research and show you how the Nasdaq disclosure research agent helped uncover those insights. The research agent is purpose built to investigate complex disclosure questions across large volumes of public company information, combining AI with Nasdaq's unique data and insights, source linked evidence, and the ability to produce finished work products. Let's look at two examples from the research. We wanted to understand how companies were evolving their risk factors in response to AI. Rather than simply searching for mentions of artificial intelligence, we asked the research agent to identify companies that added an entirely new stand alone AI risk factor in their latest 10 k. The agent leveraged Nasdaq's structured disclosure data across thousands of companies to compare current and prior year filings, help identify newly added risk factors, and distinguish genuinely new standalone AI risks from routine changes in language. The disclosure research agent uncovered 73 companies added new AI related risk factors so far in their twenty twenty six ten k's, but the agent's work doesn't have to end with the analysis. For this assignment, we also asked to turn the research into a PowerPoint presentation summarizing the findings and supporting analysis in a format that could be reviewed and shared. The result is a finished work product that takes the user from a complex research question through the underlying evidence to a presentation of the findings. Our second question required a different kind of capability: Research at scale. We wanted to understand how the largest public companies are approaching board oversight of AI. We gave the research agent a defined universe of 175 mega cap companies and asked it to analyze their proxy disclosures for evidence of AI oversight at the board level. The agent worked across that entire company set, analyzing disclosures, comparing practices, and linking its findings back to the underlying sources. The disclosure research agent found that 88 of the 175 companies or 50% disclosed board level AI oversight in twenty twenty six proxy statements. This gave us a market level view across a large company universe, turning hundreds of individual proxy disclosures into a clear picture of how AI governance is evolving in minutes. These two examples illustrate what changes when AI moves beyond answering routine questions. Teams can leverage Nasdaq's unique data and insights, conduct long running research across much larger company universes, trace findings back to their underlying sources, and turn research into finished work products like the presentation you just saw. And the disclosure research agent is one part of the broader set of Nasdaq disclosure agents. The disclosure drafting agent helps turn research and company context into review ready disclosure language while the gap analysis agent evaluates disclosures against regulatory requirements, reporting frameworks, and market practice to identify potential gaps. All three are available through Nasdaq lens, our disclosure research, benchmarking, and drafting solution for finance, legal, governance, sustainability, and risk teams. The disclosure report shows what these capabilities can uncover across the market. Nasdaq lens and the disclosure agents bring those capabilities to the disclosure questions that matter to your organization most. So, Sarah, we we just saw an example. of AI analyzing public company information at a scale that historically would have been very, very difficult to accomplish. From a CFO's perspective, what becomes possible when finance and disclosure teams have that kind of market intelligence available to them? Look. I think it's, super helpful to have more context and to be able to interrogate it and, again, to source back to where are you finding it. And so I don't think that brand new things become possible. But there are times where you would have needed to go to one of the big four, consulting fees or if it was legal, one of the law firms, to actually get an answer where you can with very good reliability because you can actually verify an audit, what they're talking about, find example. And so I think it's incredibly helpful to have. I don't think it changes what finance ends up doing, but it enables you to get your information cheaper and faster so that you can then, pay attention to other issues and to making decisions faster and to having more contacts and to make it better decisions faster rather than changing your business. And I'll just pile on to that. We call that, ownership ownership over the entire process and project. And so that you're not having to rely on, say, an outside law firm or an outside audit firm. We own we own the project. We we figure we own the data and do the analysis under the time demands. It makes us so much more efficient that we can meet, you know, short turnaround times. So, Mark, as AI becomes more more and more embedded in legal work, how does the role of the in house lawyer change? I think it's changing dramatically. So first, you, I think many of us were trained differently, and so now we're having to be retrained, to use these platforms. And, so I think, you know, that's the biggest... Just the change of behavior of of using these platforms, doing more of the work ourselves, has been the biggest... The way I've noticed the most. Can I give you my take? I have the feeling that internal legal teams are gonna have more of a role because, I would suspect that where you would, in the past, have needed to go to an external law firm, to help you. Suddenly, in your end to end ownership, you actually can get to the same quality of answer yourself. And I'm not saying that I wouldn't value, a senior legal partner for a very thoughtful analysis. But when I'm drafting a legal contract, when I'm doing something that's a bit more basic, I can actually get a lot done without, the the help that I, maybe in the past as a legal, internal function, would have gotten from external partners. I I don't know if you have thoughts on that. Yes. Absolutely. And so you're able to do... Like you said, there's a there's a cost efficiency there of not necessarily going to the outside, but you're also compressing the time it takes to do things. And so we're able to do more internally than we normally would. And, again, that ownership concept is really important to us. Like, we own the data, we own the project, we own the results, and we're not relying on a third party law firm. Now sometimes we'll go to them for confirmation if it's a if it's a significant risk issue or something like that, but you're absolutely right. So another audience question that we have is, what it... What executive... What should executives bring, to their board regarding AI? You wanna start, Mark? Sure. Sure. I'll I'll go first. So I think... First, I think it's a board's... A management's responsibility to to inform the board of how the company is embracing AI and how the company is using it and why we're using it to have a strategy, for the use. I think it's also been important, for the board to understand that because the technology is changing so rapidly, that we have the appropriate governance structure in place and, so that people are not using it in an in an improper way or risking the intellectual property, of the company. And so I think it's important for boards to know that there is a comprehensive governance structure with appropriate guardrails that has representation from, you know, different parts of the organization and that it's a real governance committee, but it's a balance. It cannot slow us down. It cannot slow the use of AI down. So it's a real balance that that has to be achieved, but I think that's really important for boards to know. Sometimes I get asked, hey, should we educate board members on how to use AI? And my response there is no. That really is their their... That should be their own responsibility in my in my opinion. We do educate them, you know, how AI works, and we educate them on on the governance around it. I agree. with everything Mark's... I I just don't... I I 100% agree with everything Mark said. And one more thing, which is we've got GenAI KPIs, which are both on the inputs and the outputs. And we are sharing a subset of that with the board. So we've got actually a dashboard that has all sorts of notions from financial performance to, HR productivity, etcetera. But we are including Gen AI KPIs, in our board reporting on a quarterly basis. And, Mike, a, an audience question for you. General purpose assistance are now in most enterprises. Where do you find you need something purpose built for the domain versus a general tool? Yeah. It's a great question and, on many software builders' minds for sure. So, you know, when I think when I think general versus purpose builder domain specific, I come back to a few really important categories that Mark and and Sarah talked about, which is data. Right? What is... Is there specific data that is domain specific? Right? And so do those tools require that? Does that domain require that? One. Two, are you collaborating with folks in a specific domain? And so does that does that, make make a big difference? And last would be that domain. Is it scientific enough? Is there something unique about it that makes it, makes it require... A requirement to have a domain specific tool? So I think it's back to data, it's back to domain specificity, and it's back to collaboration. Those those big things matter most, but, you know, I think there's so many great tools out there. Everybody's got access to so many different, options. And so my perspective is have curiosity, you know, just dive in if you're on the fence. If you're still the 15% that haven't moved, dive in, have some curiosity, and I think your mind gets gets blown when you start to, you know, use the tools and see what's possible. As Sarah mentioned, to go from half the team with the white belt to, you know, call it three quarters of the team with the white belt now, I think just shows you once your eyes are opened, there's so much potential. Absolutely. Sarah and Mark, do you have anything to add? That was. I. think I, I think Mike was spot on with it. I think the general sort of platforms can help us with communications, planning, summarizing meetings. But when you really get into the specific practice areas like intellectual property and patents, that's very domain specific and so you're gonna have to use a platform that's really trained in that area to be effective. Absolutely. And then I will maybe give one advice to the audience, which is for your data, make sure that you keep ownership of your data so that, you have more flexibility, to bring your own context, to whichever best platform or model exists that has been going through your security. And so that's very, very important. Continue to own all of the data you can own because otherwise, you're going to build dependencies, which at some point might be costly. Absolutely. And words of wisdom to bring us to the end of today's conversation. Thank you all so much. A huge thank you to Sarah Youngwood and Mark Casper for being so generous with their time and insights, and to Mike Stiller for walking us through the research. And thank you all. Thank you to the audience for your insightful questions, for your time and attention. We hope you have a great rest of the day, and thank you for joining again.