Recorded live from the CLOC Global Institute (CGI) in Chicago, this episode of CLOC Talk explores how one global legal team is turning AI ambition into real-world impact. Janessa Nelson sits down with Fabian Otto and Merve Tanner to uncover how they successfully launched an AI literacy program and governance framework that empowers legal professionals to confidently understand AI regulation while seamlessly integrating AI tools into their daily work. From practical use cases and proven change management strategies to bold predictions about agentic AI and the rise of connected legal operating systems, this forward-looking conversation is packed with actionable insights for legal leaders ready to accelerate AI adoption and shape the future of legal operations.

 Hey, everyone. Welcome back to CLOCapp Talk. We are at CLOCapp CGI Global Institute live in beautiful Chicago. This episode, we are bringing you a very hot topic that everyone is talking about everywhere at CGI, AI. Lawyers, legal ops, paralegals, we're all focused on AI adoption across the field. Many of us, including myself, have a mandate from our organization to do more AI adoption this year and in the future.

In this episode, we're gonna dig into emerging use cases and actual opportunities that are differentiating between hype and what's real. We're gonna explore the very real challenges from evolving regulations and integration issues, and we are joined by people who actually know a lot about AI and AI for lawyers and AI in the legal industry.

Today, we're joined by Fabian Otto and Merv Tanner from the Erste Group Bank. They had a session on AI for lawyers, AI by lawyers. Fabian is the legal innovation and technology lead, and Merv is the legal counsel product at/AI advisor. Thank you so much both for coming all the way from Vienna, Austria, which I know how long of a journey that was for both of you.

Let's just start with giving us a little bit of background about your journey into legal ops to let all of our listeners know. Fabian, do you wanna start? Amazing to be here. Yeah. I made the transition from an active lawyering role into legal operations or legal project management a couple of years ago, and since then, I was supporting the legal technology program we are rolling out at Erste.

Same. And Merve, how about you? Yeah. So I'm actually not really in the legal ops section. What I do is I do the AI assessment of each AI use case at the whole group, so of Erste Group. So each AI use case that needs to be assessed from a legal perspective, I'm on it, so to say. And I also advise and support the AI governance team in establishing the AI governance and also executing it.

So it's a little bit of advising them in a strategic way, but also on the other hand, really in the operative business. I think that everyone from this whole field can understand having to wear many hats and having a lot of different jobs in one. So you are always welcome in this space. Mm-hmm. So let's just give a high level summary.

So we designed eight webinar series where we invited internal and external speakers to give us their thoughts on the upcoming topic of AI and legal technology. And with that, let's say, foundation, we developed a couple of on-top activities, like a legal technology and operations program, and the setup of- Nice

a structured AI governance that is applied across the whole banking group. Yeah, so in the brainstorming phase of that literature program, we were like, "How are we gonna divide the roles we're in? Like, he does something else, I do something else, and, like, how do we put this all in one program?" And then it suddenly made sense to follow both approaches and introduce them into the part where we call AI by lawyers, meaning where the lawyers have to be aware of the new legal field, like AI law, and how to- Mm

apply the laws, but also, for example, how the AI governance looks like in the company, basically in their risk role. But the other being AI for lawyers, so how do we apply tools in order to get faster and more efficient at what we're doing? But they're both needed if you want to have an AI transformation, but they're both different.

If you put it all in one person, it's just gonna not be so good to really prosecute with. So that's why we said, "Hey, this is good idea to divide also our literacy program accordingly." Mm-hmm. Yeah. No, I think that is very relevant for the times too, and people are trying to figure out how to be literate with AI.

And especially in the legal field, I don't know exactly what- Austrian legal field looks like, but at least in here in the US, lawyers are very tech-phobic in some ways, right? Mm-hmm. And so everyone's sort of running to get into AI and AI adopt. So having sort of a guideline of how to be literate in AI, I think, is a really- Mm-hmm, mm-hmm

useful sort of- It's just also about- Yeah ... creating that awareness. Not saying every legal person must have the skills to assess an AI use case, for example. Mm-hmm, mm-hmm. Because if you're not a tech-savvy person, you're probably not be, gonna be able to understand how the technology exactly works and then how to do the assessment.

Yeah. But if that use case lands on your table and you don't know what to do with it and you don't send it to the right lawyer, so to say, that's where the actual problems happen. And so- Yeah ... this part of the program was more about telling the lawyers, "So this is a new field of law." Yeah. "Pay attention.

These are the experts you need to go to in this company." But it's not about you now have to do the assessment yourself because if you're a contracting lawyer, that's what you're gonna do, right? You're not gonna suddenly assess any- Yeah ... AI use case. Mm-hmm. But if you don't have that awareness, you're never gonna know and not gonna lead the person to the right, correct- Yeah.

Exactly. And, uh- ... direction. The underlying rationale was how are we supposed, as lawyers, to guide the AI or the bank's AI transformation if we do not use the technology ourselves? Exactly. So we thought we will definitely need this wholesome strategic approach on getting lawyers to use it so they will be in the position to actually advise about it.

Yeah. And I think you mentioned it perfectly. The real challenge right now is the, the change management and the adoption part. So we also spent some effort in structuring that- Yeah, yeah ... uh, quite well. So let's talk about AI adoption and let's use you all as the example. How have you been doing AI adoption at Erste Group Bank?

Mm-hmm. So as I mentioned, we are a pretty decentralized company. We've a headquarter in Vienna, eight major daughter banks, and then 46 saving banks in Austria. So we did a couple things. First thing was we adopted a so-called SPOC network. So a single point of contact. Okay. And then we- tried to advise the sparks who should lead and advocate the change in their teams to follow, let's say, a couple of recommendations.

The first recommendation was try to create some safe spaces around AI adoption because we're having, I don't know, ten-plus languages in the group, so all of the trainings are usually in English. English is not the primary language of most of them, so- Yeah ... these safe spaces should be designed in a way that it's on smaller scale, people coming together in their teams in a non-judgmental area where they can speak their local language and where issues are getting addressed quite promptly, so nothing, let's say, builds up around that.

The second thing was we tried to play a little bit with the gamification or use the change narrative. Because when you bring in positive emotions to change, when you frame it like we have to climb this mountain together and stuff, that actually helps. And from, let's say, a project management perspective, you really need to remove every friction you possibly can, making it as easy as possible for users doing their first steps into a new tool.

And last but not least Try to establish a knowledge-sharing culture. Mm. So if people only work in silos, and they're not sharing what they are doing, will not, let's say, m- move you as a whole organization forward. So we really try to design some formats where people could come together, exchange about what they're actually doing on working, where do you see the problems when they try to translate their day-to-day practice into AI.

So from small, let's say, exchanges to big legal tech breakfast. We did every quarter where every team shared, let's say, one use case they kind of brought us. But maybe to also take it one step back, so what's really helping is when you're even able and willing to do all what you have, like with the AI tools you have, but you have to have the AI tool first, and so our- Okay, yeah.

Yeah. Sure, yeah. So I think that's when Fabian was very, like, early on already pushing this strategy with our head of division, and we got those tools, and we have them. And we see with other companies, bigger banks as well, that they don't have these tools, but we, like, have them, and it's, I think, also important to have these tools to get going and get started.

And I, on my end, from the governance side, so as a stakeholder of the AI governance board, I was also pushed into the cold because we have an AI officer who's very passionate about innovation. So he said, "I don't want this old-fashioned governance tool. Like, people, the business owners, do not want to fill out forms of how this tool or how this use case looks like.

We need some kind of chat interface and et cetera, et cetera." And, like, I, as a stakeholder from legal side, was pushed to design and program together with the developers this tool, so it would get the information out of it. That was also my first experience with the governance team and the innovators on that side.

So yeah, you never know where it's coming from. But when it's coming, then it's good because then you get to try it out, and then you just find your way in. Yeah. This episode is sponsored by Discernus. When your team lands five hundred thousand documents in six languages, keyword search won't cut it.

Discernus is AI native document intelligence that reads the full record, maps every key entity, builds linked chronologies, and answers questions with citations back to the exact passage. Built for e-discovery scale. Deployable on-prem when privilege demands it. To learn more, visit discernus.ai. I think that that's a relevant conversation that's happening in a lot of companies, especially the larger companies where they have very real concern about AI governance.

And basically, that's how we build it step by step. You need to do an AI Act assessment, so you need to classify that use case, if it's a high risk, if it's a prohibited practice, if it's a gen AI case, and then there are legal consequences to it. But law is one part, but you need to be able to execute it in a living environment.

And so we interpreted it in, how does it work in the organization? How can we translate the law into the operative business, so to say? And that's how we step-by-step build it. And now the way it looks is we have, like, center of excellence, which is like a, the core governance team in the operative side. The use case owners, business owners, they have to submit the tickets about the AI use case, describe the purpose, the technology, target architecture, and we get it on the agenda on a Thursday, have time till Monday.

On a Monday, we meet up, and then we discuss the stakeholders, deliver their assessment, and they're good to start developing. Oh, great. Yeah. Um, if that's, you know, if, if- That's really fast, actually. Yeah, yeah. It's very fast. Yeah. It's very demanding as well. And then they develop, and before they're ready to go live, so can be in two months' time, can be in two years' time- Yeah

that's when they come back again because, of course, in development, the use case might change. They might see, oh, this is not working, but this is working. And so that's where we also check the architecture, the infrastructure. It's a very flexible environment, and that's how it's possible to develop those use cases.

And maybe it's good to mention that in highly regulated industries, you're not Designing or starting a new governance approach on the green field, so there's tons of regulations and processes already in place. Yeah. And so you need to really find a way how to integrate that on top, that you do not- Yeah

create parallel activities, and that you really try to enable everyone who is part of the governance to work as fast and as aligned as possible and with- Because people always say, like, "You're so over-regulated in a bank, like how do you manage?" But actually we're so used to being regulated that we're like, "Okay, next regulation.

Give it to me." In some ways it's turning the sort of legal industry from saying no to yes and. Yes. Yes, you can do this, but we need to put parameters, right? Yeah. So not saying no, but saying yes and, let's work together. There's a lot of people who are at CGI who are in very small legal departments. Maybe they're wanting to do AI, but they have either no budget or a very small budget.

But I would always try to start with the actual problem you want to solve and not with the dedicated tool, because the answer might be completely different if you end up with a Copilot or a Claude Pro license. That should be centered about the actual use case you're having. And usually from my experience, if you're starting with it, try to find the sweet spot where you have a use case or a set or a category of use cases that appear with, let's say, a certain degree of frequency, but ideally has a lower legal complexity level.

Because that's actually the classic quick win where you want to dive in. That's where you can get some lift off of the shoulders from your legal team, and then just take it from there, evolve. What I see a lot is when you bring in AI to legal colleagues, some of them tend to say, "Yes, now I will, let's say, get rid of the most 10 complicated percent of my job."

And actually you should do ... It's the opposite. Yeah, yeah. It's the opposite. You don't want to end up in pure frustration. Yeah. It's maybe better to start the other way around. And also I think what's really actually easy to start with where you... if you don't have a vendor that's already onboarded, with legal the good thing is if you have legal questions, like legal questions you want to clarify, you do not really have to input data that's sensitive.

Because if you have a legal question, like you could- Formulated in an abstract way and try to find out. This way you could maybe try out different tools that are maybe also free or not, but if you really want to process document, that's when you would of course- Yeah ... or customer sends data. But this is a way to get started to just get a feeling, I guess, how it works, what tools would work for you for your legal questions, and obviously it's gonna get better if it's an enterprise solution, but- So let's, if you can, let's give some practical use cases of where you all have seen really high impact in your sort of AI tools and your AI adoption.

Yeah, from my side, as I was telling you, so it's quite demanding the, how the AI governance routine works. Every Thursday we get 10 to 13 new use cases we have to assess until Monday. That is a lot. Yeah. Yeah. It's a lot. And so I realized, okay, this is a very similar assessment I have to do every time, and Fabian introduced Legora, and I've been using workflow within Legora to do the first template, use the template- Yeah

do the first assessment, and then I just work on it. And it sounds very simple, but it has helped me, like, and taken away a lot of pain I had before and provided more time for really thinking about, is this really correct? Rather than first thinking, now I have to write this sentence, and now I have to write this sentence.

Uh-huh. And this is just very simple, but- Mm-hmm ... yeah, this was h- very helpful. Taking off the sort of routine work and allowing you to do the high level sort of complex work- Mm-hmm ... is sort of part of what we want from an AI tool, right? Yeah. Yeah. Exactly. Coming back to the metrics with finding the sweet spot for...

In banking, that would be, let's say, questions you get from the retail branches that are related to, let's say, more simpler structured products like loans, saving accounts, and then you have customers who have questions. And usually you can ask, answer these questions with, let's say, the underlying contract and with a couple of internal working instructions or policies.

So if you collect the data you need, put it in a database, and then for all, let's say, the questions you're having- Yeah, yeah ... or you're receiving against that, you will get a very first, and usually very good draft on how- Mm ... to communicate the legal position in a very quick way to colleagues in the retail branch.

And then a more, let's say, sophisticated use cases, we have colleagues who are supporting loan and finance agreements with corporations, so there's a lot of contract work. You're receiving a master service agreement from a External, like a tech vendor, and you decide, okay, this time we're not gonna use our MSA, then you can quickly check, for example, is the external MSA according to the internal standards, and can you benchmark it- Yeah

and then point out, point out stuff that is not working. Also, what's always helpful is when you think of something, like you get started and you're like, "Now I really don't want to do this." I think that's where the best AI use cases That's a re- really good tip, actually. Yeah. Really good tip. Where you have to, you know, overcome that first feeling of I don't wanna get even started with that.

Yeah. And I think that's where it really helps. That's true. And I think you should never underestimate, also lawyers are tasked with non-legal task- Yeah ... from time to time. So completely depends on your role. I would say most of our leaders are using it for more, let's say, strategic tasks in sense of communicating stuff to other stakeholders in the bank, communicating stuff, for example, to external counterparts we're having.

Yeah. So in that regard, the tools are also a really great support because it helps you to translate this very complex and regulatory-driven wording into something more relatable There's a lot of stuff I think giving people specific use cases that you're doing can be really helpful to be like, "Oh, I actually wanna do that, too.

That's a good idea." So thank you both so much for sharing. It's really incredible to see what everyone's doing. Last thing before we wrap up, what do you see coming, what are you excited about in the next year or maybe five years related to AI and AI- Mm-hmm ... adoption or maybe even AI governance? I'm really curious and really, really excited about what's to come with agents and agentic AI.

I think it's just gonna be a game changer, and I think it's just gonna change the whole narrative of how we work today and how do we see our workflows also in the legal department, but also, you know, like, the whole company. And I think it's really gonna ruffle some feathers and just gonna- Oh, boy. Yeah

change... Yeah. So it's gonna be very exciting, and I think it's just gonna change a lot. Yeah. I would agree. AI will be more integrated with all of the tools you're using. Some of the, let's say, AI features will get a little bit in the back, and you will not even, let's say, see that there is actually AI behind it.

Mm-hmm. From a legal technology perspective, I think the most interesting thing will be how, let's say, more complex tasks or processes will be translated to a kind of legal operating system that- Okay. Yeah ... connects more or less all of the tools, tasks, and workflows you're having. Mm-hmm. So we will not have this classic, let's say, media breaks.

You get an email, you have to drag it into your legal assistant- ... to get something out of it. I think in one or two years of times, this will not just happen anymore, that you have to transfer data from one tool to the other- Mm-hmm ... and it will be smoothly operating. And accessibility to, to humanoid robots in a cheaper sense- Oh, yeah

so you can use them at home for- That's what we're hoping for ... all your chores. I... Y- your mouth to God's ears. I, I hope so, too. Thank you so much for joining us, and thank you for coming all the way out to CGI- Thank you ... to talk about this topic and to give us helpful, practical tips and show what is working and what's successful.

So thank you both. Thank you for the invite. Thank you. Amazing. Pleasure to be here. Yeah. Great.