Speaker 1: Excellent. Very happy to be with you today. Not going to speak too much about legal translations, don't you worry. I'm more here to talk about my innovation journey. I've been asked to explain to you how it's possible to go from being an attorney in an M&A and capital markets law firm and end up speaking in front of you at the UPFL machine learning days about reinventing legal translation with AI. And not alone, as a lawyer would be, but with a multicultural team next to me. An innovation journey, we're going to talk about how to embark, what the navigation hurdles are and what we can say being in the middle of our trip. First of all, what is the spark that you need to embark? It can ignite anywhere, believe me. As a lawyer, my job was to observe and give advice. But every time I was advising a client, I felt like standing up, taking their place and implementing the piece of advice I had just given them. I wanted to get my hands dirty and build something. So all budding entrepreneurs have their own motivation. If you, for instance, can't bear hearing that to reach excellency, you must repeat something 100 times, you think, hey, 100 times the same thing, let's automate it. Or if you're ready to sacrifice your vacation and your salary to bring your own brick to the wall, then you might be ready to hear the call of the entrepreneurial journey. But like me, where to start? Sitting in a law office. If your highest diploma is the bar exam, then it might be a good idea to learn some applied algebra. So I decided to do a master in business and administration to, so to speak, learn how to use a compass and navigate by the stars. But you don't need to do an MBA, which is quite a lot of time and a serious investment to get the necessary toolkit. In Switzerland, you can find many boot camps, incubators, accelerators, and coaching programs like the coaching program from InnoSuisse. And this was absolutely instrumental for me. Plus on top, you said you got 30,000, which was nice. And SEAD gave me 10,000 to start my journey. It feels like a real gift from the stars. But it just really gives you the mental support you need to start your journey. So I decided to bootstrap my first company, which was a classic service company. I don't advise you to do that, but at the time, it seemed like a good idea. I wanted to use what I knew about law firms to optimize their processes. Law firms are rather resistant to having their process optimized, and they preferred us to solve another issue they had, they thought was more important, legal translation. So we did what entrepreneurs call a pivot, sorry, pivot. And we became a highly specialized legal translation agency. Again, we were not offering anything new, but our understanding of the topic, the workflows, and our ability to recruit the right talent were the two secret ingredients to build a successful, again, service company. Still not advise you to do that. We worked and worked. We grew and grew steadily and happily for over 10 years. And from being a startup, we became an SME. That's when you start buying your first USM furniture, and you rediscover the meaning of the word vacation. Is this the end of the journey? Certainly not. Easy times present their own risks, and all entrepreneurs should know that calm seas precede the coming storm. Therefore, you should be on the lookout for the next big wave. For us, it was obviously, sorry, big wave, clicking problems. For us, the big wave was very obviously neural machine translation, six or seven years ago. And for other service companies today, it's obviously Gen AI. What do you do if you see the wave coming? We decided very clearly that it would be a mistake not to write it. We've decided early on to create Lex Machina, which was the first neural machine translation engine specialized in Swiss law. Who, we thought, better than our specialized linguists could build such a solution. But we had two ingredients missing. The first, and everyone who knows about AI knows that, it is more high quality data. We had loads of data. We knew where to get a lot more, but we didn't know how. No technical skills. Second, what can you do in AI if your crew is made of lawyers and linguists? On any innovation journey, these people would not get far without the engineers. And sorry to all engineers in the room, that's how AI sees you. For an SME, especially in the legal and financial industry, trying to innovate without engineers and finding the engineers is a big hurdle. How do you find them? Where do you find them? How to convince them? And how even talk to them? The solution to both our challenges, data and the team, was a Nino Suisse mentoring check with the invaluable Robert von Kober. He introduced us to the School of Engineering of Fribourg, and we've immediately connected. Together, we filed a Nino Suisse research project to develop the first highly efficient data gathering and processing pipeline for legal data in Switzerland, probably in the world. And that completely changed the DNA of our company. What you see here is not an AI generated picture that you could easily prompt by asking, what does a successful Nino Suisse research project look like? It requires a strong commitment by every specialist to really open up to the domain expertise of others and to trust them, which for a lawyer is not always easy. The resulting collaboration between lawyers and linguists led us to get three of the four magic ingredients to seize a business opportunity. First, a deep understanding of our market, its inefficiencies, and the client needs. Second, the knowledge of how technology can help you solve part of these problems, the perfect team to implement it. And we just needed the fourth, the terrible one, the funding. And although I've been asked to talk about funding, I refuse to do so in details, because like every entrepreneur, I would so much prefer to focus just on the product, the team, and the clients, and not on what often feels like this, while everyone tells you that you should look like that, and make it clear to everyone that investors are standing in line to entrust you with their money. Here again, there are really multiple initiatives at your disposal in Switzerland to help you. We've been part of the PWC RACE program and are starting now the Startup Campus Female Founder Initiative, also supported by Nino Suisse. And this doesn't give you money, but it really does give you the support, and again, this kind of kick that you need to get going. The good advice they give you is always the same, race early, always be on the lookout for more, and spend very wisely. So that's it. Once we had this last hurdle, we were able to launch this new venture, Neuron, in August 2022, with the great conviction that if it's not us who can solve legal translation, no one will ever do. Just a little bit still, bear with me about what legal translation is and how to solve it. Of course, as you can imagine, in the center of our solution, we have machine translation. But we're getting further, of course, than the DeepLs and the Googles out there, by offering a selection depending on the document, or even depending on the part of the document you're translating, to always give you the best models, or to train, or fine-tune, or prompt a model for you. And then thanks to our 16-year experience in project management, we have this absolutely instrumental part in our solution that automates what professional project managers do for their clients in translation agencies. And if you look at the whole workflow, this part brings the highest efficiency gain. And lastly, of course, mistakes. There are mistakes. There will be. Human in the loop is essential for trust, and also really for quality, let's just face it. And AI can support the human in supporting the AI. So that's our pipeline. I've been asked for the technical challenges. The first is, of course, that if you consider the speed of new LLM go-to markets and the speed of innovation, it's very important for us to build a solution that is model-agnostic and that can integrate the latest innovation as soon as possible. The second, and it's the same probably as for you, Leila, is really the language and legal culture component. So we have this jurisdictional limitation. But the best language models at the moment are mainly based on an Anglo-Saxon data, Anglo-Saxon culture. And this is an issue. We need models that are specific and that really understand the legal culture of each jurisdiction we're working for. But of course, again, we have several plans to face this issue. Then working with law firms and banks, we have to pay particular attention to data, security, and workflows. This is a big challenge. It means that when lawyers want everything to be 100% Swiss, they mean it. It means that every step in our process has to be developed both with the best models available that are maybe international models and with an alternative that we can host on our or on the client servers. So this makes every training and step a little bit more challenging, which we like, of course. And then finally, when we started with AI a couple of years ago, it was impossible to find GPUs in Switzerland that you could just rent the way you could in Europe or elsewhere. So that has been a big challenge. Swiss infrastructure, of course, there are cloud providers for GPUs, but not as many as you would think or you would hope, and not necessarily with the biggest machines. So that has been a bit of a struggle at the beginning. It's getting better, of course, since the buzz. So again, even with these technical challenges, my conclusion is that Swiss waters are not the worst to sail on. On top of Innosuisse, so many other actors in our country that make your journey easier. I would like to mention Trust Valley, who is also here today with a stand, and their whole ecosystem of data security specialists. They're really important if data sovereignty, data security matters to you. Then we're very excited about the opportunities offered by this huge computing system that has been built in Ticino, in the Alps infrastructure. And finally, and here I'm just mentioning the Swiss Legal Tech Association, but the association community in Switzerland is very strong, and the willingness to help and to collaborate is enormous. That's it. Thank you for your attention, and good luck on your own trip. Thank you very much, Paula, for the presentation.