Kamal Youssefi

#TikTok moved in Zurich in 2025. OpenAI, Google, Apple, Meta, IBM, Microsoft, NVIDIA, and Huawei are already here. Why Zurich? ETH talent plays a role, but the real reason? Switzerland is becoming the "bank" of the digital age. Once a haven for financial assets, Switzerland is now a safe house for data. TikTok’s new Zurich subsidiary isn’t managing the platform—so why is it here? The answer lies in Switzerland’s legal gray zone: outside the EU’s strict GDPR rules, yet deeply connected to global markets. Swiss banks once protected wealth from regulators. Now, Swiss servers may be doing the same for data. In a world where data is more valuable than money, Switzerland is emerging as the world’s data vault. As AI and Big Tech race toward unrestricted data flow—what happens when the world’s most valuable asset is stored in a neutral zone? If Switzerland holds the data… who really owns the future? Your thoughs? #AI #Switzerland #Data

2.611

138 Kommentare

I built a GenAI system that doesn’t hallucinate. Here is how: I worked on a project to help Swiss public institutions save hundreds hours of work a year. Here’s the problem our team tackled with Zazuko GmbH: - Swiss public institutions use graph databases to store and share data. - Only data experts can access it, because it requires writing complex database  queries. - For everyone else that knowledge is blocked. You can force everyone to learn a query language. Or you can build something that helps with it. The solution? A chatbot that lets non-technical users access data from graph databases. By just using natural language. You ask questions in plain language. But behind the scenes, there’s something more. A ChatGPT-powered engine plus RAG translates these questions into database query. The system understands data structure, and gives only relevant answers.  No hallucinations. Here’s the trick: Even if the model does hallucinate, it doesn’t matter. Hallucinations mean invalid database queries. Most of them get filtered out. Some make it through - but just return no results. Failed queries are never exposed to the user. The agent either knows the answer and gives it or it doesn’t and tells you that. So the user always gets 100% accurate results or no results at all. Never anything in between.

206

82 Kommentare

AI, developed responsibly, can improve lives. I shared this belief in Fokus’ “Future of Technology” feature—and it reflects what’s unfolding across Switzerland today. In over three decades, I’ve never experienced a technology adopted as quickly as AI. And Swiss companies are among the leaders in practical implementation: According to our 2025 Work Trend Index, 52% of Swiss companies already automate entire business processes, above the global average.   The impact is tangible. Think about the Spring ACT – Action. Compassion. Technology.'s Sophia chatbot, developed in Bern, that supports victims of domestic violence worldwide. Or think about Virtuosis AI in Lausanne: Their solution can detect burnout with 90% accuracy through voice analysis. Zurich Insurance, in turn, leverages AI for more accurate risk assessments, while EKZ applies it to optimize energy distribution.  And now we’re entering the next chapter: Agentic AI—systems that understand context, plan, and execute, always with humans maintaining control. IDC forecasts 1.3 billion agents globally by 2028, and Swiss organizations are actively shaping this transformation. We already see Frontier firms emerging in Switzerland—companies that look markedly different from those we know today. Structured around on-demand intelligence and powered by hybrid teams of humans and AI agents, these organizations scale rapidly, operate with agility, and generate value faster. Our commitment matches this momentum: four Swiss data centers, partnerships with ETH Zürich and EPFL among many others, and a goal to skill one million people in AI by 2027.  AI is most powerful when it works hand in hand with people—amplifying insight, creativity, and impact. 👉 Read the full interview (in German only) here:  https://lnkd.in/ex-Yauxp

182

10 Kommentare

Last year, I collaborated with a Microsoft Product Group on creating the Native Vector Search feature inside SQL Server. At daenet, and ACP Holding Digital AG, we have been working diligently on our AI Platform, which facilitates the implementation of #GenAI in enterprises. I am proud to announce that #SQL Server Native Vector Search is an integral part of our AI Platform. Why SQL Server? SQL Server is widely deployed, enabling us to complement our long-term cloud journey with the ability to deploy AI models in sovereign environments or maintain data sovereignty. Our solution, provides an excellent environment inside ACP Private Cloud for both sovereign-only and hybrid (#Microsoft #Azure) deployments of AI applications—projects we have been developing over the past years. This is where SQL Server Native Vector Search proves invaluable. As part of this effort, I authored an article (1) demonstrating how .NET Enterprise Developers can leverage this feature. All samples related to this article are available in the official Azure Sample repository (2) on #GitHub. References: (1) Article: https://lnkd.in/e5FPw_Je (2) Samples: https://lnkd.in/e7HMxr7z ACP Gruppe If you want to do it the right way, think about how to deal with your data -First things First. Welcome, SQL Server 2025!

65

6 Kommentare