Ignite clients : Emily Prince (Directrice de l’IA groupe, LSEG) et OpenAI

[music] >> Hi everyone. My name is Nikolai Skobo. I help lead Open Eyes European go-to-market team. Um, as you heard from Matt earlier, for this next session, we're going to go ahead and invite uh a few of the executives from some of the financial services organizations that we work with uh across Europe and and hear from them. And to start, I'd like to welcome Emily Prince uh together with me to share a bit about the London Stock Exchange's journey with AI. >> [applause] >> Thank you for joining us. >> Thanks for having me. It's great to be here. >> Well, very good. Um, so Emily is the group head of analytics and AI at the London Stock Exchange Group. Um FTSE 100 uh global financial markets uh infrastructure and data provider, as I'm sure many of you know, supporting 44,000 customers, I think, globally with 26,000 employees. >> Yeah, it keeps us busy. >> Keeps you busy uh across 170 markets. Maybe just to start, could you tell us a bit about your role and and maybe LSEG's role in the market as well? >> Yeah, happy to. So um starting with LSEG, so LSEG's role is really to provide financial information, trusted financial information. And uh sometimes that looks like underlying data, pricing, but also market models. So you can think of risk models and many others. Uh the other side of what we do is um really to provide that unbiased picture and also services. So from clearing to the exchange, which many people associate uh with the group. Um from myself directly, I'm I have two hats really that I wear. One is in how we use AI inside LSEG. So from the tools that we use to the way that we scale and go from a thousand flowers blooming into something that really is with scalable impact. Um and to the other side, actually, how we service our customers as well uh using AI safely. >> You have something called Al Seg everywhere. Um, an AI strategy. I I'd love to hear a little bit about it and for you for you to sort of tell the group how it's evolved, maybe how it came about. Uh, and how you're sort of disseminating it into the organization. >> Yeah, happy to. So, with Al Seg everywhere, it was it was something that we believed in very strongly or and still do of course. When people think about AI and what is going to look best for their organization, that looks like different things. It sometimes looks like different tools, but also recognition of just how fast this space is evolving. And so, what we wanted to do and together with Open AI, what we made available were things like our MCP. So, model context protocol and it allowed us to really unlock the enormous quantity of data and our just taking our data and analytics business alone, it's over 33 petabytes of data and bringing that into the hands of customers very directly. So, people working in chat GPT can ask those questions, build the reports that we were seeing a little bit earlier and actually not just ask generic questions, but ask very tangible questions grounded in trusted data, but also with those deterministic financial models at the same time. So, Al Seg everywhere was a belief that regardless of where you're working, what tools you you need to have that anchor in terms of trusted information to make sure you're going beyond just general summary of information with AI to something that is a genuine transformation in the way that you work. >> And and this applies internally at Al Seg. Are you thinking about this externally as well with a similar concept? >> Absolutely. So, it's interesting. So, when when we started to really think about how do we go from these definite experiments that we have in the group to something that is genuinely scalable? That was one of the really big things that we wanted to focus on. Um, and the the challenge that we have is uh, Al Seg had multiple acquisitions, lots of different data model and so on and so forth. So, we had to think about how do we bring this together in a way that we could unlock all of this advantage genuinely at scale, both internally, but also for the benefit of customers. We heavily leaned into our API strategy and with that MCP as that unlock in terms of being able to use a lot of these great tools over the top and that that was a true game changer for the way that we work. >> Together, we've talked a bit about, you know, the the evolution of of AI. I'm I'm curious to hear sort of how you're thinking strategically about AI today versus maybe just just a year ago. You know, what are you what are you investing in now maybe as well to set yourself up for the the next phase to come? >> Yeah, so I don't think I say evaluation >> No. >> [laughter] >> I think every day I say the probably evaluation framework multiple times a day, so that that count is definitely going up and I think in that as we've scaled more and more and gone from these kind of single contained spikes of expert to something that is genuinely scaling, we've had to really lean on on how do we make sure that we're upholding that quality and achieving our intended outcome. And that came in a few places. One, we had to be really purposeful on what problem are we solving? How do we get success from teams in finance to teams in marketing to teams in our product to teams in engineering, but also from a client perspective to the teams in IBD to the teams in management to the teams the analyst research teams. So, it's really recognizing all of those different persona groups and then being really clear on what does success look like. And for us, that went beyond model selection and the harness that we use into different applications, but also heavily lent into what does our evaluation framework like and how can we make sure that that's really done at scale across all of these and with that we can really run at speed. >> How have you sort of seen the the adoption across the organization? Are there certain aspects of the the team, the strategy, the culture that you've seen sort of shift over the last 12 months? >> Yeah, it's a really interesting one. So, I think you get this question a lot around like what are the skills, what are the and actually I find it to be hugely culturally rooted in terms of attitude more than anything else. It's just those people that are really leaning in and experimenting and creating and we just see these incredible leaps of the people that are just open-minded, experimenting, playing and the kind of that that moment of like, "Oh my And for me personally, I've had quite a few of those. I I remember from my time at various different banks that I used to work in doing horrendous reconciliation in Excel and debugging macros and all sorts of horrible horrible things. And what we can achieve now and the fact I can talk to my spreadsheet and I have is kind of radical in terms of the way that we can work and I feel like we're only at the beginning of actually realizing just how powerful that is. And that would be one thing if we were talking about generic information, but coupled with the fact that we were actually exposed this source with the MCP, means you're actually reaching out with turnkey access to all of that trusted data. There was no data onboarding for 6 months. no big program. It's turnkey access to all of that content generatively into a spreadsheet that looks like something that would have taken me several hours historically. So, slightly depressing as well as very exciting. >> Exciting and and sort of presents the opportunity, I think probably across the whole organization. Uh You you touched on analysts earlier. Uh helping transform how analysts work has been a a priority and and sort of initiative from you and the team. What was the core problem for analysts before AI? >> Yeah, so I think with analysts in particular, you know, it's really common that you're talking about multitude of different data sources, structured and unstructured sources, and candidly, we all have a certain amount of time in our day, and we need to go pick up our children, and there's lots of other things that we need to do. And so, over time, we've kind of conditioned ourselves to only pull from certain pieces of information. But it as an analyst, you really want everything. You really want everything at your disposal, and ideally all of those slightly orthogonal insights are really going to differentiate your analysis and give you something that is of a bit of an edge. The question is, where is that going to come from, and is it genuinely trusted source, and are you going to have to spend a lot of time pre-processing it before you can actually use it? So, for me, it's the fact that you can be so much more bountiful in terms of the sources you're pulling from. You can uphold that standard. You can embed with things like scale, a lot of the standards, the policies, the preferences, the biases as part of that at scale. And you can be very kind of fast in terms of iterating through different cycles where you can really provide the feedback in in real time. You're actually iterating through what would have taken potentially hours or days previously. >> Yeah, and and you all were obviously one of the first financial services organization to launch an application on ChatGPT, sort of creating a surface for analysts and your your customers to to leverage it in other services and platforms. You and the team are moving incredibly fast. Uh and as we articulated earlier, covering, you know, 40-plus thousand customers across countries all over the world, keeping up with the technology and the innovation is super challenging. How how are you thinking about sort of balancing the rapid adoption of the technology and the pace over the court of control that you naturally need given the industry you're in? >> Yeah, it's uh interesting, Pete. So, I think for us, we had to try and is a constant battle. I'll I'll I'll say up front. >> Yeah. >> But try to be quite grounded on what are the no regret pieces and um that distinction was quite important. So, for us in what we do, the no regret things were the work that we did around API and with that the MCP because we saw that information unlock of the difference between we were whether we working with relevant intelligence from the perspective of the job to be done from our standpoint or we were just working with something extremely powerful but also very generic. So, that was a really important one and the universality in model selection that goes with that, the harness that we use around it has really proven to be really powerful for us. On the other side, what we're doing is and together with you guys is really constantly pushing each other. So, we spend a lot of time Nico with your team kind of pushing you on problems that are heavily kind of financial service specific on a lot of the things that we worry about in terms of deterministic practices, workflows around risk, around regulation and how do we get this balance because there are inherent problems that we've had in the industry for a long time. This is not a kind of panacea and sometimes a lot of the frontier things that we're talking about don't necessarily directly apply until we go that extra mile. So, I feel like it's that constant very active partnership where we're pushing each other but with that being very very open-minded. So, the whenever we're looking at 55 or and it's where it's having that really open-minded approach to what could this look like? What does the next frontier and how do we make sure that we're really agile in terms of the way that we lean into that. The other thing I'll say is there's a big cultural piece in the way of working for us as well in that you don't have these like long development cycles anymore. They're very short, they're very fast. It is everyone in the tent together building, iterating, building it and it's really fast. So, in the past, uh from throughout my career, candidly, you've got your PRD, you do your intake, you've got these kind of long cycles in terms of process, and we're just seeing that radically shift. And with that shift, we can be much more agile in terms of some of those frontier opportunities. >> And we have to be an agile partner to you, right? So, that sort of speed doesn't compromise trust or the governance model that you already have. Um Maybe along that line, it's like what what have you found that has been the most challenging or the hardest as you've scaled some of these solutions in the technology? >> So, I think maybe I'll take you this in a couple of fronts. So, you mentioned about governance. Um I think we wanted to So, we actually developed our responsible AI principles and with that an expansive governance framework across everything that we do quite early on, and that was very really much by design because we wanted to create the scaffolding in the way that we could allow the teams to innovate at scale. We didn't want to like handcuff people, but we wanted to create a safe framework where people could really ideate properly. >> When did you start thinking about this? >> Uh 2 years ago. We had it before, but it really came into we realized the scale of innovation that we were going to be looking at, and that this had to be institutionalized in a much greater way. What we didn't do was create a whole new set of rules. What we did is we looked at everything in terms of the end-to-end way that we build and service, and actually looked at all of the things that we needed to adapt with the interest of not constraining the business, but enabling it and embedding it as part of the general workflow. Um so, I think that wasn't necessarily a problem, but definitely something we had to think very carefully about, and we keep looking at. So, as we get new models, as we get new harnesses, as we start to see compression of workflows, which I'll argue is one of the most um complex things to look at. So, as we start to think about you've got your traditional characters in building something today, as we start to see workflows change and how that they're built and you go from 10 people to maybe one or two people solving that problem in a fraction of the time. We have to make sure that we're embedding all of the governance and process within that and that that's going to be something I think that we will be continually looking at to make sure that we're getting right. The bigger challenge I think is more cultural. So making sure that we people are excited, but there's often a fear that sits alongside that excitement and there's often a lot of miscommunication in terms of AI. So for me it's trying to get people safely engaged in a way that they're not left behind, but they can really engage and use it in a way that actually solves real problems for them. And I still think we're really early in that journey. >> Are you doing anything culturally with that sort of excitement internally? >> Uh we're doing a lot. So first of all it's about enablement with uh chat GPT, making sure they've got the MCV, so they've got all of the tools as one. Second thing is all about education, making sure people have those programs, have that education. We're getting much more kind of activated in terms of rather than just awareness and learning, but really like hands-on builds now in terms so that it really becomes very tangible. I love the finance examples earlier. I think that is exactly what it should look like where it is that really hands-on and accountable way of saying I'm not solving generic problems, but it's really specific problems for me, for my team and then it's something we can really invest in. >> Maybe we can ask Sarah if you can join our finance hackathon which is for our CFO. Um before we wrap based on sort of your learnings over the past few years, what what advice would you sort of impart on on your peers? Sort of what you've learned along the journey. >> I don't know if it's advice, but I think like I mean it's definitely a lean in moment. Yeah. Um and I think it is probably the most fascinating and extraordinary moment to be in this industry and when I look at all of the horrible things I mean I've been doing so 20 years and the horrible processes and ways of working and some of the like there's a lot of heavy-handed things in finance not necessarily by design but by evolution and the opportunity that we all have at this moment in time to really like to openly challenge and self-disrupt and say is that something that we absolutely have to have in that way and is there a true regulatory reason it has to be done like that or is this a moment where we can actually reinvent it because if it is something we can look at differently gee the insights that you can actually get and the opportunity to differentiate and have your organization just fly is just there and there's not a blueprint for what it looks like we talk a lot about this there is no there is no book you can read or blueprint or precedence from we're all in it we're all learning we're all creating and that feels for me truly special and exciting. >> It does for us too. We keep we keep reinventing the the blueprint together. Yeah. Um thank you so much for joining us and sharing. >> [applause]