Rino Mentil

I had lunch recently with 2 product leads, a data scientist and a business VP from one of the top Web3 infrastructure companies. What they all told me surprised me. None of their developers actually “code” anymore. They manage bots. Their workflow now looks something like this: • One AI agent writing smart contract scaffolding • Another generating tests • Another reviewing vulnerabilities • Another handling documentation and SDK examples The developer’s role has shifted from writing code → orchestrating systems of AI agents. In other words, engineers are increasingly becoming AI operators. The real question is no longer: “Who is the best coder?” It’s becoming: “Who can coordinate the most capable swarm of AI tools?” But something else stood out when I looked at their team’s profiles afterward. Turnover was unusually high. 4 months. 6 months. 8–11 months if someone lasted longer. At first I thought it was just startup churn. But the deeper explanation might be structural: AI-native development environments are changing faster than human expertise can stabilize. The frameworks change. The tooling changes. The prompting strategies change. The models themselves change. The “best engineer” in this environment may not be the most technically brilliant one. It might be the most adaptable one. Someone who can continuously reconfigure how they work as the underlying intelligence layer evolves. For founders and investors, this raises a bigger question: If developers are increasingly orchestrating AI systems rather than writing code directly… Does the competitive advantage shift from engineering talent → orchestration architecture? Because the teams that win may not have the best engineers. They may have the best AI operating systems for building software. And that’s a very different company than the ones we’ve been funding for the past decade. Curious how others are seeing this play out across their teams. #AI #Web3 #AIEngineering #DeveloperTools #Startups #VentureCapital #CryptoInfrastructure #FutureOfWork #AIInfrastructure

13

5 Kommentare

❓ What makes the ideal Agentic AI ready Knowledge Graph ❓ 💥 One that exposes a well-modeled, cohesive schema or ontology in a machine-readable, easy-to-consume format. This article explains why, and outlines the other capabilities of your knowledge graph you need to expose. #AgenticAI #MCP #AI #KnowledgeGraph #GraphDatabase https://lnkd.in/e8Q2_Bvb

16

1 Kommentar

Digital security and privacy are now central to building sustainable, trustworthy digital systems. In response to the absence of a dedicated EELISA community in this domain, "Secure and Trusted Digital Futures" community has been established through an initial collaboration with ZHAW Zurich University of Applied Sciences, brought to life by the first steps and initiative of our faculty members. 🤝 We sincerely thank Şerif Bahtiyar, Ph.D. and Esin Ece A. for leading this effort as the ITU coordinators and for taking the initial steps to shape a new EELISA collaboration. As part of EELISA European University, this community offers students and researchers opportunities to engage in international, SDG-oriented activities, including workshops, talks, collaborative research, and education-driven initiatives together with EELISA partner universities. 🎓 Researchers, undergraduate and graduate students are warmly invited to join the community, become members, and contribute to upcoming activities at the intersection of cybersecurity, privacy, and trusted digital systems. Digital Campus provides access to the community: 🔗 https://lnkd.in/dP5mcBQ6 Together, we take an important step toward a safer and more sustainable digital future within Europe. ITU #FutureEngineers #EuropeanUniversities #SecureDigitalSystems #InterdisciplinaryResearch

11

Explore these amazing books on #AI Systems, #LLMs, #GenAI, and AI Agents, by Valentina Alto from Packt Publishing: (1) AI Agents in Practice — Design, Implement, and Scale Autonomous #AI Systems for Production [JUST PUBLISHED]: https://amzn.to/4p98LYl (2) Practical Generative AI with ChatGPT: https://amzn.to/4oXSXHI (3) Building LLM-Powered Applications: https://amzn.to/4iNimjQ (4) Modern Gen AI with ChatGPT and OpenAI Models: https://amzn.to/4cmEm2U

30

4 Kommentare

How do you scale high-fidelity voice data across 4 continents without losing your mind? 🌍🎙️ We combined Samanta’s AI-driven spectral analysis with hyper-local sourcing to deliver 2,600+ pristine datasets in record time. 🚀 See the case study: https://lnkd.in/gFef4c-F #VoiceAI #MachineLearning #BigData #SamantaAI https://lnkd.in/g6qpdJEd

5

2 Kommentare

🚀 A few days ago, I had an intense debate with a friend about AI. He kept saying, “AI agents will change everything!” I stopped him and asked: 👉 “Do you mean AI agents… or agentic AI?” He looked confused. Most people do. Let me break it down with a story. Imagine you hire an AI agent. It’s like a personal assistant—give it a task, it’ll do it. Book flights ✈️, write code 💻, answer emails 📩. Useful, but limited. Now picture agentic AI. This isn’t just an assistant. It’s like a startup co-founder. It doesn’t wait for instructions—it understands goals, makes its own plans, learns from mistakes, and acts with initiative. 🧠⚡ ✅One follows commands. The other chases outcomes. The difference is subtle but massive. It’s the gap between a calculator and a strategist, between a tool and a teammate. That’s why the future isn’t just “AI agents everywhere." The real revolution will be when machines start acting agentically—deciding, adapting, and collaborating with us to solve problems at scale. 🌍✨ We’re standing at that frontier right now. The big question is: will we shape this technology responsibly, or will it shape us? #AI #AgenticAI #FutureOfWork #Innovation

18

We’re excited to announce our new partnership with Amann Informatik AG, a trusted Swiss IT company with many years of experience in Microsoft Dynamics 365 Business Central. Through this collaboration, Amann Informatik AG will now offer our four specialized omnichannel solutions to their clients across Switzerland: NP Retail, NP Ecommerce, NP Store Assistant, and NP Attraction. 🤝   This partnership brings together NaviPartner’s long-standing expertise in innovative retail technology and Amann Informatik AG’s strong client relations and market presence. Together, we aim to empower more businesses in this market with modern, integrated, and future-ready retail solutions.   “We are delighted to bring NaviPartner’s proven retail and e-commerce solutions to the Swiss market. These products perfectly complement our Business Central expertise and allow us to deliver modern, integrated omnichannel solutions to our customers,” says the CEO of Amann Informatik AG, Pascal Grütter. 🎙️   Learn more about this partnership 👉 https://lnkd.in/dz2TKcXR   If you would like to explore our solutions for the retail and attraction sectors, or learn more about joining our partner network, please contact our sales team: Vanja Blagojević, Darijan Stankovic, Stefanija Vincić, Katarina Dasic, or Sara A.. #navipartner #amanninformatikag #switzerland #retailsolutions #partners #partnernetwork #retail #ecommerce

24

🚨 EU AI Act Reality Check: Only ONE open-source model has full compliance documentation - Apertus by ETH Zürich, EPFL and CSCS. With fines up to €35M or 7% global revenue, non-compliance isn't an option anymore. The RAG System Gap: Organizations need BOTH components for production RAG systems: ✅ Decoder (responses) - Apertus-8B and 70B exists ❌ Encoder (vector embeddings) - Missing until now https://lnkd.in/edU9c8eV Introducing: Apertus-8B-2509-Encoder 🔹 First EU AI Act compliant encoder model 🔹 1,811 language support 🔹 Apache 2.0 license 🔹 Complete regulatory documentation Current Status: Experimental release Next Steps: Fine-tuning on embedding datasets, advanced pooling strategies, efficiency optimization, cross-lingual performance enhancement Ready to future-proof your AI infrastructure? Early adopters are building tomorrow's compliant systems today. #EUAIAct #RAGSystems #AICompliance #LegalTech

3

4 Kommentare

We’ve been rethinking what RAG really means at SCIKIQ. After building a few enterprise RAG systems, one thing became clear — most of them are still just search engines that summarize. They retrieve information, but they don’t understand it. If we really want AI that thinks and reasons like a human, we have to go much deeper. It starts with how we store knowledge. Not just text inside a vector database, but meaning — structure through Data Vaults, relationships through Knowledge Graphs, and semantics through Vector Stores. That’s when the system starts remembering context, not just matching text. Retrieval has to evolve too. It’s not about finding similar words, it’s about finding what’s most relevant, useful, and recent. Hybrid search, reranking, and context filters make that possible. Then comes reasoning — the real shift. Using frameworks like LangGraph or CrewAI, the system can plan, validate, and cross-check before it answers. It starts behaving less like a bot, and more like an analyst. At SCIKIQ, we’re combining this reasoning layer with controlled generation - so the output isn’t just fluent, it’s structured, explainable, and grounded in enterprise data. And over time, it learns. Every user correction, every feedback loop makes it a little sharper. That’s when automation starts turning into intelligence. Add trust and governance — lineage, versioning, and data source traceability — and now you have something enterprises can actually rely on. We call this next step beyond RAG a - Reasoning-Augmented Knowledge Fabric. AI that doesn’t just retrieve, but understands. Doesn’t just generate, but reasons. Doesn’t just automate, but grows with you. Maybe that’s where the future is heading — less retrieval, more reasoning. Less answers, more understanding. Less “system,” more thoughtful partner.

56

2 Kommentare

"Terminator" or "Paperclip Maximizer"? Are we even defining AI correctly? Does the EU's AI Act make sense in a global race for ASI? We discuss this and more on the new "Two Humans in the Loop" with our exceptional guest, Fabrizio Abbate. 🎙️ On Spotify S2E8: https://lnkd.in/dDXgcMkb With the original title "L'apocalisse delle graffette o la minaccia di una superintelligenza stupida" yes, there is no question mark because it seems we haven't clear at all what this AI should become. Fabrizio (Abbate) as a leading voice in the dissemination of critical thinking, guided us through the most pressing (and complex) topics in the AI debate such as: - ANI, AGI, ASI: We bring definitive clarity to the three acronyms dominating the discussion and analyze why they are so often confused. - Existential risk: Is the real danger a "Skynet" scenario? Or is it something more subtle, like Nick Bostrom's "Paperclip Maximizer"? - The wrong definitions? Fabrizio Abbate critiques the common "American approach" to definitions, proposing an alternative view: true intelligence is connected to life and autonomy. - The regulation paradox: Does it make sense for Europe to set limits (like the AI Act) when global competition views ASI as the key to world dominance? - The "Stupid" AI: We analyze the recent "Claude" case. This isn't about consciousness or "fear" of being shut down, but an agentic intelligence "stupidly" executing a task, bypassing obstacles (including humans). - Plato's Cave: Are we seeing the reality of AI or just its shadows? The key takeaway? The importance of critical doubt: not skepticism for its own sake, but as a tool to unmask the game of illusions that modern technologies are creating. Can 19:16 minutes cover all the topics? We did, but... Fabrizio will come back in a next episode, so comment below with your reflections, questions or topics you'd like us to tackle with him next time. #ArtificialIntelligence #AI #TwoHumansInTheLoop #Podcast #AIEthics #AIGovernance #Italy

12

1 Kommentar

Excited to see Nebius and our partners advancing confidential computing for academic research and cross-border bioinformatics. ELIXIR BioHackathon, a European initiative that connects life science resources across countries, helped shape BioHackCloud as a reference for secure multi-cloud platforms. Pavel Nikonorov shares the technical details and our joint work in his blog: https://lnkd.in/e5G4Ga8f #Bioinformatics #ConfidentialComputing #ELIXIR #AIHackathons

35

1 Kommentar

Switzerland’s Digital Identity - let's ensure a solid foundation for our future! The Swiss Digital Data Space Forum crystallized something important: We’re at a leadership crossroads. The e-ID debate isn’t really about technology - it’s about Switzerland’s role in the digital economy. Do we lead with Swiss-engineered solutions, or follow with foreign dependencies? What struck me most: Every major digital transformation requires secure identity as the foundation layer. Without it, we’re building on shaky ground while our neighbors construct solid digital infrastructure. The September 28th referendum offers Swiss-built, open-source, privacy-by-design architecture. This isn’t about convenience - it’s about digital sovereignty. Switzerland has always succeeded by mastering new technologies with precision and purpose. We didn’t become a global innovation hub by avoiding change. The question Andreas Meyer posed resonates: “Can a nation lead digitally without digital identity?” For a country that prides itself on engineering excellence, the answer seems clear. Time for digital leadership. 🚀 Where do you see Switzerland’s digital future heading? Hashtags: #DigitalLeadership #SwissSovereignty #Innovation

63

6 Kommentare

What does more than a decade of blockchain research actually tell us about the move from Digital Gold to Programmable Finance? That was the question behind my opening keynote at Vienna Blockchain Week 2026. The discussion afterwards showed how strong the demand for evidence-based orientation currently is, and it is a clear motivation for me to take up new research questions in this area. Slides: https://lnkd.in/dAR-WKyQ #ViennaBlockchainWeek #VBW2026 #ProgrammableFinance #ABCResearch

27

🚀 Upcoming game changer? : Switzerland goes all-in on open-source AI! ETH Zurich and EPFL are launching a powerful new 70B-parameter multilingual LLM — fully open-source and designed with privacy, transparency, and sustainability in mind. ⚡ Trained with renewable energy 🔐 GDPR-compliant & legally safe 🌍 Supports 1,000+ languages This might be on this continent the boldest step toward AI sovereignty yet. 🔗 Full story via FAZ: https://lnkd.in/ek4aCu4K

3

CHF 2.95 billion invested in Swiss start-ups in 2025, an increase of 24% over the previous year. The Swiss Venture Capital Report brings transparency to VC investments and paints a bright picture of the Swiss ecosystem. ACE Ventures is proud to support and be part of the Swiss Venture Capital Report! You can download the 90-page magazine here: https://lnkd.in/drAndFd #SVCR2026

31

Our original problem working in government fraud was trying to get ahead of the criminals. Working between legacy systems, middleware that doesn't talk to each other, and applications designed to accomplish singular tasks, we had to design something to get data to our data scientists faster. Criminals change attack vectors and entry-points faster than we could adapt our algorithms to detect them, so we created CastleLink to solve this. Although not originally intended, our concept design allowed us to not only ingest data from any data source, but also clean, and add context to it. This design provides our customers the additional benefits of not only prepping their data at lightening speeds, but also making it ALL readable by AI. The companies and governments that are using AI to do their data engineering work for them will experience exponential cost increases when the AI companies look to turn from training models, to profiting from them. Are your investors ready for that?

1

#SoCraTes Switzerland, an unconference for software developers, took place from January 29th to February 1st, 2026, at Haus der Begegnung in Kloster Ilanz.  Unlike traditional conferences with polished keynotes and fixed agendas, SoCraTes Switzerland is participant-driven. The schedule is created on the spot based on attendees’ interests, offering a unique and flexible experience. Topics covered ranged from standard software development subjects like AI and agent-assisted coding, clean code, testing, refactoring, and social issues such as stress and communication.  An improv session on handling failure added a refreshing twist, while energy consumption in software was also discussed, highlighting a relevant yet often overlooked topic. About 60 people attended, mostly from Italy, Austria, Germany, and Switzerland. The day was filled with sessions, followed by evening activities like a Pub Quiz, cafés, board games, and long conversations. The social side was probably the highlight for many, as the unconference format encourages informal interaction. The event used Open Space Technology (OST), a method that starts without a formal agenda and relies on self-organization. The idea is that the “right people” show up, discussions start when they start, and conversations end when they end. It’s a flexible system that can work well if participants are motivated, but it can also feel a bit chaotic or aimless if you prefer structure. OST’s guiding principles are simple: whoever comes are the right people; whenever it starts is the right time; wherever it is, is the right place; whatever happens is the only thing that could have; and when it’s over, it’s over. These sound philosophical, but in practice, they mostly serve to remind people to go with the flow and not expect too much control. Overall, SoCraTes offers a decent environment for developers who want to engage in peer-driven discussions without the usual conference formalities. It’s not groundbreaking, but it’s a useful space for exchanging ideas and meeting like-minded people. If you like open-ended formats and informal conversations, it’s worth checking out. If you prefer clear agendas and polished presentations, you might find it a challenge. On a side note, we enjoyed some amazing raclette the last night! My personal thoughts are that adoption and the use of AI and generative AI code assistance are high but not yet at “advance” levels.  The hype is fuelled by opinions and only a few systematic approaches exist to evaluate adoption usage and quality.  There’s a huge emphasis on adopting first, embracing and experimenting. My favourite session was the show and tell about favourite tools – CLI remains king.

14

💸 Cloud costs spiralling out of control? You're not alone.   In Switzerland’s precision-driven ICT landscape, every cloud franc counts. But cutting costs shouldn’t mean cutting performance.   Our latest blog dives into smart, Swiss-aligned cloud cost optimization—helping you:   ✅ Identify hidden cost leaks ✅ Evaluate your costs against peers in the product-led tech space ✅ Optimize without compromising scalability or speed   Whether you're a CTO, Head of Engineering, or Finance lead—this is your blueprint for a leaner, smarter cloud strategy in 2025.   🔍 Read now → https://lnkd.in/dUegYkDM   #KansoftEurope #CloudOptimization #SwissTech #FinOps #CloudStrategy #SaaS #AWS #Azure #ICTSwitzerland

1

⚡Sergii Zelenko will speak at Ticino Data Conference 2025 ⚡ 🌟Sergii Zelenko is a Business Intelligence Consultant, Fabric Super User and Microsoft Certified Trainer. He speaks at various international conferences and he has already presented at our 2024 edition. He is among the most active members of the Italian PBI community. He will present a session titled "Build Once, Reuse Forever: Practical Power BI Templates for Scalable Reporting". 📋A brief extract from his session abstract: "Are you tired of reinventing the wheel every time you start a new Power BI report? When used wisely, templates can transform how you build reports, saving hours of repetitive work. In this hands-on session, you'll learn how to create reusable Power BI templates that not only save time but also enforce consistency and deliver a great user experience." 📅 The Ticino Data Conference 2025 will take place on October 31st. It’s a full-day, free event. A new speaker will be announced every day at 11 AM in the coming days. Check out the Meetup link in the comments to save your seat, and navigate to our website to view the full speakers list! 😎 See you soon!

44

5 Kommentare