Johannes Maron

I’m excited to share that I’ve been invited to speak at this year’s workshop of the German Interest Group in Operating Systems (https://lnkd.in/d2WGUQM3), hosted by Heinrich Heine University Düsseldorf on 26/27 February. In my talk, “Using AI to Fight GenAI Fraud at Scale,” I’ll explore how generative AI is reshaping the fraud landscape — from synthetic identity creation to industrial-scale deepfake attacks — and what it takes to defend against it effectively. Beyond detection models, I’ll dive into the engineering side: what enterprise-ready, cloud-native architectures must look like to handle millions of verifications per day, process billions of events per month, and scale dynamically while maintaining reliability and observability. Looking forward to discussing the intersection of AI and large-scale systems with the operating systems community.

56

Did you know you can use mittwald's hosted AI models (currently in beta) for coding assistance in both VS Code and JetBrains IDEs? Just pushed a bit of documentation (read more at https://lnkd.in/gPKSUHVV) that explains how to configure both JetBrains AI Assistant as well as Continue (for VS Code, or JetBrains IDEs) and Cline (also VS Code) to use open-source models like Qwen3-Coder-30B-Instruct via the mittwald API. All models are hosted on german servers, and no user data is stored (let alone used for training purposes).

29

Last Thursday, I held my first session at the "AI Software Engineering Bootcamp" 🤖 of Digitale Leute School. With AI, the capabilities and responsibilities of engineers are widening. Engineers can - and need to be - closer to the customers and to the product. That is why the first module is about Product Management. 🚀 In my session about "Stakeholder Management and Storytelling", we dove deep into: → Reading the room: How to address stakeholders in different kinds of companies (business-led vs. tech-led vs. product-led) → Prioritizing with spine: Prioritization frameworks that work → Working backwards: Using PR/FAQs to align stakeholders around big bets by turning complex products into simple stories → Saying "no" without burning bridges: How to do your most important job without risking relationships Despite being three hours in the evening, I was very happy about the energy and engagement of the participants. 🔋 My next session is coming up in Feb and will be about "Coding with AI responsibly". Looking forward!

62

8 Comments

Excited to share this Claude Code Meetup happening in Munich this Monday, February 23rd. Three talks covering real-world applications, from AI-powered tools in healthcare to consulting workflows and multi-agent systems. What makes this interesting is seeing how teams are actually building with Claude Code in production environments. The range of use cases shows how versatile these tools have become, from early-stage products like halbzeit.ai to strategic consulting work and complex multi-agent architectures. Great to see the Munich AI community coming together to share practical insights. Thanks to Andreas Cappell and Accenture for hosting, and to the organizing team for putting this together. If you're in Munich and working with AI tools or curious about real implementation stories, this could be worth your time. Sign up: https://luma.com/nlfxyfut

9

1 Comment

TYPO3 CMS Migration framework In 2025 I worked on a big migration and upgrade project. It was a website based on TYPO3 CMS, that needed a major technical overhaul. During the upgrade process I migrated over 100 content elements from DCE to Content Blocks and Container. On the TYPO3camp Hamburg 2025 I presented my approach to a handful of people. Some of them approach me afterwards asking about a similar migration for Flux based content elements. Matthias Krappitz put in some effort and found 4 companies including his own to sponsor the development of the Flux to Content Blocks migration. The result is not only a Flux to Content Blocks migration. It is an extensible migration framework for content element migration. The core uses Content Type Providers, that provide normalized configuration and data of the source content elements, Content Type Builders, that build the configuration for the target content type and Record Data Migrators, that take the source data and return the data structure for the new content type. Based on the structure it is possible to migrate from one content element strategy to another. It would even be possible to migrate plugins or core content elements with this, if someone built content type providers for that. Currently there are extension with Content Type Providers for DCE and Flux available, as well as Content Type Builders for Content Blocks and Container: * https://lnkd.in/eg-pirJD: Migrator core * https://lnkd.in/ezGYBXnw: DCE Content Type Provider * https://lnkd.in/eRkUDCPx: Flux Content Type Provider * https://lnkd.in/etgSH_qZ: Content Blocks Content Type Builder * https://lnkd.in/etiDADVs: Container Content Type Builder If you think this could be relevant for you, please start with reading the documentation (README) of the migrator core, as it describes the core concepts and components. If any questions arise please message me here on LinkedIn or send me an email to thorben@webcoast.dk. This the first version, which still needs some testing. So if you're interested, please give it a try an provide feedback. A big thank you goes to aemka, Apart Design Studio, Homepage Helden GmbH and Henrik Ziegenhain for sponsoring the development. Another thank you goes to Nikita Hovratov (https://lnkd.in/efiDk59H) for his work on Content Blocks and b13 GmbH for their container extension. #typo3 #cms #dce #flux #contentblocks #container #sponsor #community #opensource

8

When a small German health provider found itself with a growing demand for scalable and secure technologies it wanted to find the right cloud solution. Not only did Azure deliver on all fronts, but it also helped reduce IT costs despite a higher number of configurations.

6

From idea to production in 10 hours. I wanted to see how fast you can actually ship something useful with AI coding in 2025. Not a demo, an actual product people can use. I have built www.smartzeugnis.de - generates German employment references. If you know German corporate culture, you know these are weirdly complicated. They use coded language where "zur Zufriedenheit" sounds positive but actually means "barely adequate." Perfect problem for AI. Stack: Codex + Claude Code, Next.js, Mistral AI, Hetzner. 10 hours of work spread over 2 days. What I learned: The breakthrough wasn't coding faster. It was meta-prompting – I used Claude to write prompts FOR Claude Code, then manually validated the output. This loop of prompt → build → validate is ridiculously fast. You stop thinking in code and start thinking in features. I know this sounds like hype, but most companies are still developing like it's 2019. Better tools, same process. That gap is only getting wider. Try it if you speak German: smartzeugnis.de and let me know your feedback. Curious how others are actually using AI for development – are you coding with assistance or orchestrating AI to build for you?

73

25 Comments

“Efficiency and transformation are really changing the way we are working.” That’s how Helmut Schuller, Managing Director at @SCHULLER & Company GmbH, summed up the momentum we’re seeing across industries. As one of our long-standing partners, he joined us at #AVEVAWeek 2025 to share how joint projects with AVEVA are helping customers modernize operations and unlock new potential. 🤝 Hear more about how strong partnerships drive real-world innovation—with measurable impact.

33

1 Comment

What are AI Multi Agent Systems? And How can they help corporates save millions ? 𝙊𝙣𝙘𝙚 𝙟𝙪𝙨𝙩 𝙙𝙧𝙚𝙖𝙢𝙨... # 20 🎉 Meet NISAR Autonomy GmbH : pioneering multi-agent AI for engineering process automation! 📌 Website: https://nisar.ai/ 👩‍💻 𝗧𝗘𝗖𝗛𝗡𝗢𝗟𝗢𝗚𝗬 🤖 𝘜𝘵𝘪𝘭𝘪𝘻𝘦𝘴 𝘔𝘶𝘭𝘵𝘪‐𝘈𝘨𝘦𝘯𝘵 𝘚𝘺𝘴𝘵𝘦𝘮𝘴 (𝘔𝘈𝘚). What are they? ▪ MAS is very similar to a business organization in industry, where different employees or experts with different expertise and experience work together. ▪ The individual AI Agents have different behaviors and use different types of AI or classical data engineering algorithms. ➔ MAS is digital team of multiple specialized AI-Agents to solve a multi-step problems with high accuracy and consistency. –with the human in-the-loop. How does this compare to ChatBots and AI Studios? ▪ ChatBots are 1-step solutions: they cannot execute complex strategies. Also, their context is limited, they can fetch data once. ▪ AI Studios offer different versions of Vector-RAG. This gives good answers from simple data like one or few PDF. For the industry we on the other hand need complex data analysis and aggregation from complex data sources. ➔ MAS provides higher accuracy and completeness from complex data 💰 𝗘𝗖𝗢𝗡𝗢𝗠𝗜𝗖𝗦 💸 Chatbots automate ~40% of knowledge/document tasks; Nisar’s MAS boosts efficiency to 80–90%+. 📈 𝗧𝗥𝗔𝗖𝗧𝗜𝗢𝗡 𝗮𝗻𝗱 𝗨𝗦𝗘 𝗖𝗔𝗦𝗘𝗦  Nisar have multiple use cases as its solution cna be applied to any technical Document-Heavy process.  Let's make it concrete, one of the most mature application is Manufacturing root cause analysis. ⏱️ Traditional root‑cause analysis takes 1–4 weeks; MAS compresses this to minutes, allowing senior experts to focus on critical tasks. 👷‍♀️ A 7‑agent system works interactively: scoping, strategizing, processing time‑series data, generating diagrams, and delivering results—all for rapid insights. 💥 Outcome: dramatically reduced downtime and expert hours, preserving production profitability. 𝗡𝗶𝘀𝗮𝗿 𝗨𝗦𝗣𝘀: 🏭 On‑premise deployment reduces GPU costs and secures enterprise data. 📈 Strong scalability via modular agents and hybrid knowledge bases—easily adapts to new domains, data types, and legacy systems. Flexibility with multimodal data and legacy (IT) systems and existing processes 💬 𝗪𝗮𝗻𝘁 𝘁𝗵𝗲𝗶𝗿 𝘁𝗲𝗰𝗵𝗻𝗶𝗰𝗮𝗹 𝗯𝗿𝗼𝗰𝗵𝘂𝗿𝗲𝘀—𝗰𝗼𝘃𝗲𝗿𝗶𝗻𝗴 𝗥𝗙𝗜 𝗮𝗻𝗮𝗹𝘆𝘀𝗶𝘀, 𝗰𝗼𝗻𝘁𝗿𝗮𝗰𝘁 𝗿𝗲𝘃𝗶𝗲𝘄, 𝘃𝗲𝗵𝗶𝗰𝗹𝗲‐𝗯𝘂𝘀 𝗺𝗮𝗽𝗽𝗶𝗻𝗴, 𝗮𝗻𝗱 𝗺𝗼𝗿𝗲? 👉 Or do you want to know more AI startups validated by industry leaders? Reach out!

22

2 Comments

if you think #Rust is still a niche programming language only for hobbiest, you need to catch up on the news. Ubuntu is replacing the `sudo` implementation with a Rust version. You cannot go more mainstream and security critical than this!

31

1 Comment

’m convinced that the convergence of APIs and AI is not just a trend—it’s inevitable. Whether it’s in the context of MCP or more traditional REST APIs, the potential is huge. One platform that exemplifies this is #IBM #WatsonXOrchestrate, which opens up some incredible possibilities for innovation. I’m currently working on an article that dives deeper into these opportunities, so stay tuned! In the meantime, if you’re interested in exploring how APIs and AI are shaping the future, I highly recommend checking out the IBM #APIDays in Munich! 😊 #AI #API #TechInnovation #IBMWatson #FutureOfTech

6

We are thrilled to announce a significant strategic milestone for #naoo. Today, we are substantially expanding our software development and AI capabilities with the addition of four exceptionally experienced and well-established former IBM experts.

20

1 Comment

NOA stands for Next Open Automation from KEB Automation KG With Uwe Huber, Head of HMI and IIoT, and the CEO of the Austrian KEB, Markus Weissensteiner, I had another interview with one of the providers of open Linux-based automation platforms. If you look at the statements made in the interview, it is hard to imagine that there will be any well-known automation providers left in the future who will continue to rely on closed systems. The trend is clearly moving in the direction of Linux and open standards. KEB Automation is a good example of this. My interviewees also explained in the interview why openness not only involves a new form of competition, but also very great advantages and simplifications. It is well worth reading. The providers from the market overview are: Bosch Rexroth with ctrlX AUTOMATION, FLECS Technologies with FLECS, German Edge Cloud with ONCITE DPS, Hilscher Gesellschaft für Systemautomation mbH with netFIELD, KEB Automation KG with NOA, KEBA Industrial Automation with Kemro X, Lenze with Lenze NUPANO, Phoenix Contact with PLCnext Technology, SALZ Automation GmbH with SALZ Controller, TTTech Digital Solutions with Ubique, TTTech Industrial with Nerve, WAGO with WAGO OS and WAGO ctrlX OS, and Weidmüller Deutschland with u-OS and easyConnect. The public will be hearing a lot more about this sector in the near future. It embodies the opportunity that our economy has thanks to the innovative strength of the industry in German-speaking countries. Only here is this step towards a new, digital level of automation currently taking place. And this industry will network and make itself heard. Stay tuned. Find the link to the interview in the first comment.

5

1 Comment

𝗔 𝘀𝗶𝗺𝗽𝗹𝗲 𝗵𝗼𝘁𝗲𝗹 𝗶𝗱𝗲𝗮 𝘁𝗵𝗮𝘁 𝗾𝘂𝗶𝗲𝘁𝗹𝘆 𝘀𝗼𝗹𝘃𝗲𝗱 𝗮 𝘀𝗰𝗮𝗹𝗮𝗯𝗶𝗹𝗶𝘁𝘆 𝗽𝗿𝗼𝗯𝗹𝗲𝗺 ⸻ I recently saw a hotel quietly solve a scaling problem… and most people didn’t even notice. ⸻ I was staying at a cozy mountain resort. Cold weather, beautiful views, and a restaurant with a big outdoor terrace. But since it was still winter, everyone was eating inside. Which meant one thing. Breakfast rush. Timing was 7–10 AM. Reality? Almost everyone showed up around 9–10. You can guess what that looks like. Crowded tables. Slow service. People waiting around. But this place felt… surprisingly smooth. ⸻ Later I realized why. Before check-in, they had sent a simple message: “Please choose your preferred breakfast slot.” • 7–8 • 8–9 • 9–10 We picked 8–9. And the next morning, without thinking twice… we showed up exactly then. No one forced us. But we still followed it. ⸻ There was one more subtle detail. Small notes around the restaurant: “If you’re planning to come between 9–10, please expect some delay.” Not a warning. Just setting expectations. ⸻ And that’s when it clicked. They didn’t try to increase capacity. They shaped behavior. They didn’t scale the system. They scaled human behavior. • Spread demand before peak • Let users self-regulate • Made delays feel acceptable All without adding complexity. ⸻ This is system design in real life. We often jump to scaling: More servers. More infra. More cost. But sometimes the better question is: Can I influence the load before it hits the system? ⸻ Because the best systems don’t just handle traffic. They influence it. #SystemDesign #Scalability #Engineering #DistributedSystems #ProductThinking

7

This Thursday at apidays Munich, codecentric AG and PolyAPI are hosting a workshop that brings integration back into the hands of developers. Code First, Integrate Fast: The PolyAPI Workshop is designed for developers and architects who want to connect clean integration with real code. We will explore how to accelerate integrations without giving up control or clarity. You, as participants will work hands-on with the Poly Tree AI assistant to discover functions, copy code snippets, create and deploy server functions, and execute them both via UI and in code. This is where code-first meets maintainable integration. At codecentric, we see real value in how PolyAPI uses AI to rethink iPaaS. It is not just about automating integration, it is about empowering teams to move faster and stay aligned with modern software practices.

15

My colleague Richard Nordsieck had the opportunity to present at the recent #VDMA webinar “AI solutions for mechanical engineering – data strategy.” I think that his idea to use the Asset Administration Shell (AAS) as data provider for AI use-cases is particularly interesting. Using AAS as a foundation for structuring knowledge and data — enriched with semantic information and accessible via a well-defined standard API — shows how existing industrial standards can serve as an enabler for knowledge management and AI use cases. It’s a great example of how AAS can go beyond interoperability in manufacturing and become a technological fundament for data-driven innovation. Thanks to Richard and all contributors for sharing their perspectives! #AAS #DigitalTwin #AI #DataStrategy #KnowledgeManagement #Interoperability #VDMA #Mnestix #XITASO

3

Together with the team from pacemaker.ai by thyssenkrupp we’re bringing a challenge that sits right at the intersection of AI, data, and real-world decision making. No theoretical exercises. No toy problems. Just a very real question that many organizations are struggling with today. Curious what the question will be? Come to see us at the CHEFTREFF event! We are very curious to see how participants approach it and what ideas emerge over the course of the hackathon. If you enjoy building things that turn complex information into meaningful outcomes — this should be a fun one.

27

Yesterday I had the pleasure of speaking at DecompileD Conference in Dresden about **When software products get old** . 🎤 Since transitioning from Co-Founder & CTO at a startup with zero product history to Head of Product at SQL Projekt AG I've been navigating a completely different challenge: 28 years of product history, a loyal customer base, and a market that keeps moving. The core tension: **Stability & Continuity** vs. **Innovation & Transformation.** Four approaches we use at TRANSCONNECT to balance both sides: 1. **Broad Communication** — build trust through transparency, early and often 2. **Core Stability + Product Expansion** — protect existing workflows, introduce innovation as opt-in 3. **Platform Strategy** — APIs & SDKs that let innovators build on a stable foundation 4. **Clear Deprecation Strategy** — no surprises, long timelines, clear migration paths Thanks to the DecompileD Conference team and everyone who came up to chat afterwards. If you're working on a product with a long history — I'd love to hear how you're handling it. 👇 #ProductStrategy #SoftwareProduct #DecompileD #TRANSCONNECT #SQLProjektAG

42