Karakun

We’ll be at JCON Europe 2026 in Cologne 🇩🇪 From April 21–23, you can meet us on-site as a sponsor – and we’re looking forward to connecting with the community. At our booth, we’ll be talking about topics that are currently shaping software engineering: → Artificial Intelligence  → Cyber Resilience Act We’re also running short interviews to hear how these topics show up in real-world development. And of course: stickers, a Karakun-branded "Spitzgugge" with Basel treats – and good conversations. 🎤 Talk by Tamari Gogebashvili  “From a Workshop to a Wake-Up Call: My Journey into Accessibility”  Thursday, April 23 · 12:55–13:15 · Cinema 6 If you’re attending #JCON Europe: come by and say hi 👋 More info: https://lnkd.in/dxuFWfT3 #JCON2026 #SoftwareEngineering #Accessibility #AI

Elasticsearch 7 has reached its EOL – maintenance ended in April 2025, support in January 2026. In our experience, many teams underestimate the effort - something we’ve seen firsthand. Because Elasticsearch 7 → 8 is not a simple upgrade. It’s a complex migration. So what does that mean in practice? With Elasticsearch 8, core concepts have changed: ✅ The Java client has been fundamentally redesigned  ✅ Security is enabled by default  ✅ APIs, mappings and templates behave differently Delaying the migration increases cost, risk and operational instability. In his latest article on the Karakun #DeveloperHub, our expert Jatin Gangajaliya shares how to approach a migration from 7.17 to 8.19 in a pragmatic way – including concrete code examples, common pitfalls and lessons learned from real-world projects. 👉 Read the guide: https://lnkd.in/dDJ4DgVr #Elasticsearch #SoftwareEngineering #TechLeadership #Java #SecurityByDesign

AI in engineering is hyped. But most of it is still “blind”. In automotive and aerospace, we’ve mastered automation through simulation and MBSE. But when it comes to real intelligence, we hit a wall. Why? Because data without context is just noise for an AI. To move toward true Agentic AI, systems must not only process data. They need to understand the engineering logic behind it. This is where semantic data models like FDX come into play. Without them, your AI agents are flying blind. At the prostep Symposium 2026, we’re diving deep into this: 🎤 No Semantics, No Intelligence: Why Agentic AI Remains Blind Without FDX 👤 Mike Mannion 📅 April 14 | 14:55 | Auditorium 3 🤝 Let’s debate: Visit us at Booth 06 to see how we’re bridging the gap between raw data and engineering intelligence. Michael Baumann and the team are ready for the tough questions. Do you think LLMs can handle engineering data without a semantic layer? Let’s discuss in the comments. 👇 #AgenticAI #MBSE #EngineeringData #DigitalTwin #prostepivip prostep ivip Association

💻 Laptop. Idee. Community. Der Rest passiert im Hackergarten. Beim Hackergarten treffen sich Entwickler:innen, Architekt:innen und Tech-Enthusiast:innen, um gemeinsam an Open-Source-Projekten zu arbeiten. Kein Pitch. Kein Sales. Kein Deadline-Druck. Dafür: ✅ gemeinsam coden ✅ voneinander lernen ✅ Ideen ausprobieren ✅ Wissen teilen Ob du regelmässig zu Open Source beiträgst oder an einer neuen Idee tüftelst – im Hackergarten geht es um gemeinsamen Austausch und das Bauen an echten Projekten. 📅 18. März 2026 ⏰ 18:00 – ca. 21:00 Uhr 📍 Karakun AG, Elisabethenanlage 25, Basel 👉 Anmeldung: https://lnkd.in/eJySp4tP Mit dabei sind Hackergarten-Urgesteine wie François Martin, Ixchel Ruiz und Andres Almiray. 🍕 Pizza & Getränke gehen wie immer auf uns. #Hackergarten #OpenSource #DeveloperCommunity

Have you ever set an ambitious goal — and suddenly realised how big the challenge is? With the European Accessibility Act approaching, accessibility can feel exactly like that — like standing at the start of a marathon. The distance looks overwhelming. So many teams never start. But accessibility doesn’t begin with a complete redesign. It starts with small steps: 1️⃣ Learn the basics (WCAG, guidelines, examples) 2️⃣ Start with small improvements in your code 3️⃣ Share what you learn with your team Accessibility isn’t a sprint — it’s a marathon. In our new Developer Hub article, Cindy Hürlimann explains how teams can start their accessibility journey step by step. 👉 https://lnkd.in/dimzi23w What helped your team take the first step towards accessibility? #Accessibility #InclusiveDesign #UX

If you cannot test it reliably, you cannot scale it responsibly. 💥 As LLM-based systems move from experimentation into production, engineering standards face a new challenge: The same input may produce different outputs. Traditional pass/fail logic was built for deterministic systems. AI-driven architectures require a different approach to quality. So how do we bring engineering discipline into probabilistic systems? On March 3rd in Basel, Karakun Senior Software Engineer Mike Mannion will explore this question at Java User Group Switzerland. In his talk, “Banishing the flaky LLM text – Testing non-deterministic systems with PUnit”, he shows how statistical testing techniques can extend familiar JUnit practices — helping teams bring structure and confidence to inherently non-deterministic systems. A relevant discussion for everyone who takes software quality seriously, especially when AI becomes part of core architecture. 📆 03 March 2026 🕔 18:15 - 19:15 📍 Markthalle Basel More info and registration at https://lnkd.in/dGfd_JM6. #SoftwareEngineering #EngineeringExcellence #SoftwareQuality #LLM #Java

The biggest risk in AI integration is not the model. It’s the lack of architectural discipline. 🧠 AI functionality is easy to demonstrate. Integrating it reliably into existing systems without disrupting operations or triggering large modernization programs is where things become complex. A new article by Johannes Porzelt on our Developer Hub takes a closer look at this architectural challenge. Using our #HIBU platform as an example, he outlines how AI can be introduced as a controlled layer in front of existing applications – with clear system boundaries, validation mechanisms, and testability built in. What this means for IT decision-makers: ✅ Existing investments remain protected ✅ Risks are deliberately isolated ✅ AI becomes manageable instead of experimental No black-box integration. No forced modernization. 👉 The full article is now available on our #DeveloperHub: https://lnkd.in/dDaUuSdP If you are assessing how AI should fit into your existing architecture, we are happy to discuss the architectural implications. #AI #SoftwareArchitecture #EnterpriseAI

AI-powered search promises faster access to knowledge and more efficient processes. In practice, however, three questions quickly arise: 1️⃣ Where does the information actually come from? 2️⃣ Which data is processed outside your defined system boundaries? 3️⃣ Can results be traced, explained and audited? These questions are not about limiting innovation but about making AI systems operational in real-world environments. AI search does not have to be a black box. With retrieval-based approaches, clearly defined system boundaries and transparent data flows, organisations can combine performance with transparency. We recently shared a deeper dive into architectural approaches and practical considerations: https://lnkd.in/dq_Hetuu How are you approaching AI in your enterprise search landscape? #AIStrategy #DigitalSovereignty #EnterpriseAI #AIArchitecture