Silo AI

As a next step on our journey, Silo AI is excited to announce that we’ve joined forces with AMD. Silo AI and AMD have an aligned mission to create a global leader that will shape the future of computing. Please find below a link to the press release.

The automotive industry is evolving, driven by software and AI technologies that enhance vehicle safety, performance, and user experience. Our latest blog post delves into how AI is reshaping in-cabin environments and addresses the key challenges and strategies in AI-driven automotive systems. 🔹 Challenges in AI-Driven Systems: As we shift towards Software-Defined Vehicles (SDVs), the industry faces significant challenges such as regulatory compliance, stringent safety standards, and ensuring robust data security and transparency. 🔹 Transitioning to SDVs: The move aims to reduce ECUs and wiring by adopting more centralized architectures. This requires OEMs to improve their software development capabilities, thereby reducing manufacturing costs and simplifying system updates. 🔹 Ensuring Reliable AI Systems: It is crucial to collect representative datasets for driver monitoring and manage sensitive data in compliance with regulations like the EU AI Act. 🔹 To overcome these challenges, OEMs should focus on: ➡️ Unified Data Management ➡️ Data Anonymization ➡️ Modular Data Architecture ➡️ Regulatory Compliance By tackling these challenges, the automotive industry can leverage AI to enhance vehicle safety, performance, and user experience. Learn more about the future of AI in the automotive industry in our blog, link in comments. #AutomotiveIndustry #SoftwareDefinedVehicles #InCabinExperience

Our CEO & co-founder Peter Sarlin highlights concerns related to European AI, tech & competitiveness in a recent WIRED piece: “We have a very good basis in Europe in terms of data, compute, and talent. But that’s not enough; you need distribution. And that happens through software products, usually software platforms.” As Marc Andreessen's 2011 statement, "software is eating the world," remains relevant, AI is accelerating this trend. For Europe to truly create value with AI, it is imperative to develop and scale sizable software product companies within the continent. Without these, Europe risks falling behind in the global AI race. Currently, only two of Europe's ten most valuable public corporations are tech companies (SAP and ASML), compared to six out of the top ten in the US. Notably, NVIDIA and Microsoft, the two largest US tech companies, are each worth over 15 times more than SAP. Thank you Peter Guest and WIRED for the interview and for coverage on AI sovereignty. Link to the article in the comments.

Fine-tuned models perform better than general purpose base models. With fine-tuned models, companies can deploy smaller models with equal or better results than with base models, while staying true to their brand and tone of voice. The SiloGen LLM SaaS platform provides everything needed for LLM deployment in one platform, including fine-tuning. Learn more about the benefits of fine-tuning below, and visit our site to get started with your own LLM: https://hubs.li/Q02CBKDb0

Partnerships play a crucial role on our path to create value as a private AI lab. We bring competitive advantage to product R&D by building AI-driven solutions and products. And it's natural to partner with top European AI companies: Aleph Alpha, with a GenAI Operating System, and Mistral AI, with frontier AI models. Thank you Financial Times' TNW for the interview and for covering our last weeks' partnerships: “If you follow European AI news, you may have read that Finland’s Silo AI and Germany’s Aleph Alpha recently established a partnership to deliver “sovereign AI.” It would seem whoever is responsible for partnerships at Silo is keeping busy, as the company today announced yet another collaboration, this time with investor darling Mistral AI.” In line with this, our CEO and co-founder Peter Sarlin states: “It’s a great pleasure to partner with a global leader in frontier AI models. We’ve always had an aligned mission with Mistral AI, to build world-class AI in Europe.” We appreciate TNW’s Linnea and Thomas for highlighting this development! Link to both articles in the comments:

As a private AI lab, we're thrilled to announce a partnership with leading frontier AI model builder Mistral AI, to drive value for enterprise customers of both companies. This partnership aims to combine our platform and our team’s comprehensive end-to-end capabilities, with Mistral's advanced AI models to provide more value to enterprises across Europe and beyond. There’s a growing demand for innovative AI solutions, offering businesses cutting-edge tools to boost efficiency and drive innovation. This partnership leverages both companies' strengths to deliver top-tier, customized AI products across various sectors. Arthur Mensch, Mistral AI CEO & co-founder: “We’re pleased to address enterprise customers together. Silo AI is also integrating our premier models into their platform, a move that significantly enhances Europe's AI capabilities.” “I believe that Mistral AI’s models, combined with Silo AI’s platform and extensive experience with large-scale enterprises, is a strong combination in the rapidly evolving AI market,” continues Peter Sarlin, Silo AI CEO and co-founder. This aligns well with Silo AI proven track record with large-scale enterprises, such as Allianz, Philips, Rolls-Royce and Unilever, as well as Mistral AI’s experience with BNP Paribas, Orange, and CMA CGM, among others. Read more about the partnership in the joint press release; link in the comments.

Meet Risto Luukkonen, AI and LLM expert at Silo AI. Risto is an avid researcher with experience in training LLMs in highly distributed setups across thousands of GPUs, and in training several multi-billion parameter LLMs from scratch both at Silo AI and as a researcher at TurkuNLP at the Turun yliopisto - University of Turku. With his proven experience in building LLMs and GPT-like architectures, for example, on the LUMI supercomputer, Risto has contributed to multiple low-resource language papers and worked on e.g. FinGPT, Poro and Viking, Silo AI’s most recently published model family. Continuing to live by his commitment to openness in research and large language models, Risto is now contributing to Silo AI’s and TurkuNLP’s next-in-line model family covering all official EU languages, including also English and programming languages. Learn more about Risto and his research in our blog, link in comments. #AIforPeople #Silopolis

Silo AI is releasing the second multilingual model natively trained in all Nordic languages, Viking 13B. Viking 13B is a larger and more capable version of the previously released Viking 7B. It is similarly sensitive to local values and cultures and evidence of a new customized open source training framework for the LUMI supercomputer. Viking 13B was trained together with University of Turku’s research group TurkuNLP. Read more through the link in the comments.

Powered by LLMs, the user experience and intuitiveness of a growing number of products and services are improving. Higher expectations from users, brought on by these new technologies, are not limited to products and services of companies, but to all digital interactions, including public services. The bar to staying competitive has clearly been raised. In our work with the YTK Unemployment Fund we are providing them with a fine-tuned LLM built on top of our open source base models. This ensures that they have a model that is proficient in Finnish, Swedish, and English, and that knows the correct terminology while also abiding to YTK’s tone of voice. “We can’t afford for public services to fall behind in an environment of rapidly evolving technologies. Public services impact the daily lives of a vast amount of people. For us at YTK, ensuring that the LLMs we deploy are reliable and transparent is paramount. Having models such as Silo AI’s Poro and Viking, that understand our native languages is a crucial component for our success as we aim to create implementations that help us serve our customers better. While we are bold in our AI projects, we stay grounded by ensuring success through careful planning and execution." says Satu Koskinen, CIO of YTK Unemployment Fund. If you are interested in learning more about the benefits of having your own, fine-tuned LLM, visit our website for more information: https://hubs.li/Q02C-78w0

Silo AI CEO Peter Sarlin had the honor to contribute to the ELLIS Institute Tübingen opening ceremony, together with prominent speakers like Larissa Holzki and Profs. Yann LeCun, Bernhard Schölkopf, Max Welling, Neil Lawrence etc. On the same occasion we also announced our partnership with co-located Cyber Valley.  “We’re very happy to partner with the leading center for excellence in AI and modern robotics in Europe, with partners from both academia and industry. Silo AI’s existing research co-operation with the professors Frank Hutter and Matthias Bethge at the co-located University of Tuebingen and ELLIS Institute Tübingen on open LLMs provides a good basis for collaboration among Cyber Valley’s industrial partners Amazon, BMW Group, IAV GmbH, Mercedes-Benz AG, Porsche AG, Bosch, and ZF Group”, says Peter Sarlin, CEO and co-founder of Silo AI. Rebecca C. Reisch, Managing Director of Cyber Valley, adds: "This alliance represents a remarkable turning point, not just for Cyber Valley and Silo AI, but for multilingual AI development. It resonates with our goal to assist in fostering an AI ecosystem that respects cultural and ethical values that we as the Cyber Valley Community stand for. We are excited to explore this and other opportunities with Europe’s leading private AI lab." See our blog for more, link in the comments.