 Hi, everyone. Can you hear me well? OK. Hello, I'm Elias, the CEO and founder of CodeSphere. And let's talk about how to accelerate innovation in the cloud with AI. CodeSphere is one platform for development and deployment. And before we talk about what this actually means, let's talk about how development teams work today. So they write the code. They test it. That fails. They write the code again. They do a code review. That fails. They do it again. We don't have to play this game until the end. But you can see it's a sequential process where you discover issues late with exponential costs. That's also a complex setup with constant maintenance. It requires expensive engineers, called DevOps engineers, to keep the whole system running all the time. What our platform does, it combines agile development with agile deployment. And this allows you to paralyze your development process, getting you to go to market six times faster, at 100% less people costs for DevOps engineers, at 90% less infrastructure costs, and 70% less operational costs. But today we talk about a very specific use case, which is how to use CodeSphere for hosting open source generative AI models. Generative AI unlocking the power with CodeSphere. Let's talk about what revolution large language models actually are. And we have AI since about 15 years. But we had to have small models trained by machine learning engineers on very specialized use cases. That was created by AI specialists, and it needed a lot of training and lots of data cleaning before you could get it done. Now, since JetGPT is released and open source AI models are released, if you have a model trained on your own data, you can actually use it for many different use cases without having to retrain it. This allows large scale optimization across your company without having to hire ML engineers, without having to wait for months to be able to get any process optimized. I now hand over to Roman. No, no. In a second I'll hand over to Roman. This is actually the wrong slides. OK, sorry. Let's talk about the typical costs and benefits of self-hosting such a model. The big benefit is that you own the IP. It can be trained on unlimited amounts of data, and it can work with unlimited amounts of data by a fine tuning, and it's typically faster as well. The cons are if you hosted yourself, it has extensive costs, like typically a few A100 GPUs that cost about 15K a month. It requires expensive DevOps engineers again. It requires ML ops engineers again. And it takes about six months to ramp up until the infrastructure is set up. Now I hand over to Roman. Thank you, Alice. So how does CodeSphere solve the cost issue? We've introduced a new patent technology called Reactive Inference, which allows us, instead of running AI models constantly, to start them in milliseconds only when necessary. Through Reactive Inference, we're able to reduce the cost of running low traffic AI models by more than 90%, allowing for quick scaling without having to book resources in advance. How does CodeSphere fix the complexity issue? We make things as simple as possible, automating infrastructure setup and maintenance so that you don't have to, meaning that you can start hosting your own AI model with CodeSphere in less than two minutes. How does CodeSphere make your organization faster? We introduce bundles and flat rates to make your cloud expenses predictable and controllable, making you never run out of resources even during gig times. And this is how it all comes together. What used to cost a monthly lease of a Porsche now costs less than a cup of coffee. In other words, running a single 70B Lama 2 model used to cost 15,000 euro per month with CodeSphere, you can run 40 70B Lama 2 models for less than 1,000 euro per month. And now, ladies and gentlemen, I would like to pass it to one of our customers, Johannes, from Kern AI. Please play the video. Hi, my name is Johannes Hutter. I'm co-founder and CEO of Kern AI. And today, I want to share with you how we make use of CodeSphere. At Kern AI, we mainly work with insurance companies and companies from the financial industries. And we help them to build internal tools using Gen AI to face very knowledge-intensive tasks. And as you can imagine, insurance companies process a lot of very confidential information. Let me make a bit more tangible what applications can be built using Kern AI. And I want you to imagine that you're working at the customer services department of an insurance company. You can imagine insurance companies have incredibly intangible and immaterial products. It's very difficult to grasp them. And it's typically the case that, for instance, B2B customer has a question about the insurance that a customer service cannot directly answer this question, but instead need to look up information in the internal documents. For instance, our customer Markel has a lot of internal documents that describe how and when a customer is insured. And when a customer has a question, the customer services looks into those documents trying to find the right piece of information and then gets back to the client. Of course, in the era of JGPT and large language models, there are much better ways. And this is a very popular application for internal tools to simply ask a question and then to respond to the large language model. And as you can imagine, this is not guesswork, but instead it's based on the facts from the internal documents. So there are many, many applications and also obstacles that we face every day at Kearn.ai. And one of them is that we need to face very, very confidential information. And this is precisely where Codesphere helps Kearn.ai because we not only work with large language models such as HRGPT, but we also make use of self-hosted large language models that we host directly via Codesphere. This way, we can offer our clients, for instance, for cases where they process more confidential information, much more secure and self-hosted large language models. Obviously, our customer wins. We win and Codesphere wins. And so this is a great application to technology that Codesphere offers. In these cases for it, we worked, for example, with the large German bank, where we are the application platform. We worked with the large German supermarket chain, where we allow plain text data analytics via the large language models. And this is how development of the future looks like. Developers have self-service. They can use no-code managed services, AI models, and custom code, and don't have to ask the IT department and don't have to ask their managers to get it done. They have parallelized feedback cycles for our platform, and they have a decentralized organization. If you want to know more, we have a booth over at the startup area. Thank you for the talk. So.