Olivier Nano

AI-generated code can behave in unexpected ways if developers don't thoroughly review it! Dynatrace CTO Bernd Greifender explains six key reasons why AI-generated code could make it more difficult to maintain the reliability and security of software over the long term

2

At Elemental Cognition, we’re transforming #GenAI solutions using #ApacheKafka and Confluent. In this The New Stack article, I shared how our AI platform uses Kafka and Confluent Cloud to stream and vet data in real-time before the EC Reasoning Engine applies formal logic for transparent, accurate solutions. Check out the article for a deep dive into the best practices we’ve learned for building reliable GenAI apps. https://lnkd.in/eU5RupZu

41

4 Kommentare

🔍 "Does AI truly make a difference?" Johannes Koch from Hewlett Packard Enterprise raised this vital question in his recent Handelsblatt piece, challenging the conventional wisdom around AI's role in driving competitive advantage. Generative AI holds immense transformative power – but it also pushes us toward a paradox. While productivity gains are the often-touted benefits, they risk becoming commonplace as tools proliferate, industry practices settle, and "Best Practices" become the norm. True advantage, then, lies not in the betterment but in the difference of AI application. Here's where we must think big – and think differently. 🌍 As with cloud computing, we’re witnessing a gravitational pull toward a handful of major "plug-and-play" AI providers. This could lead to an ecosystem heavily concentrated in the hands of a few, and Europe needs to ask itself: are we merely defending "digital sovereignty," or are we aiming to unlock real, unique value? In my view, the path forward is in cultivating open, hybrid AI architectures that prevent dependency on any single vendor and reduce costly switching barriers. Building unique customer value means prioritizing the kind of data exclusivity that traditional methods can't simply replicate – data from production, internal knowledge bases, or a proprietary customer base. 🌐 Independence isn’t just choosing European providers over American ones. It’s about constructing a flexible, resilient AI strategy that enables substitution when the market shifts, prices rise, or technology evolves. But the real bottleneck isn't capital or tech expertis – it's leadership and imagination. We need to view data and AI as integral production assets, akin to raw materials or equipment, and foster the same drive for innovation and continuous improvement. And this is what we do at proALPHA. I urge leaders to think of data as a competitive asset, not a byproduct. The companies that can break free from a one-size-fits-all AI model and tailor applications that directly enhance customer experience are the ones who will truly harness AI's transformative potential. If we do this collectively, we can avoid a market dominated by homogeneity and instead cultivate a rich, diverse AI ecosystem. Let’s reimagine #AI in Europe. 🚀 #proALPHA

13

With MongoDB, we help engineering leaders get off the modernization hamster wheel and embrace predictable, optimised spending that fuels their most ambitious initiatives. With powerful features like automated cluster sizing, on-demand pausing, and granular spend visibility, MongoDB Atlas helps you right-size your data infrastructure and take complete control of your IT costs. Perfect way to break free from the constraints of economic uncertainty and get easier budget approvals. This blog helps you understand how MongoDB Atlas can help you take better control of your data costs. #MongoDB #cloud #innovation

29

1 Kommentar

New Episode: 260: Amazon Dispatches AWS CEO Adam Selipsky with Prime 2-day delivery Episode 260 of The Cloud Pod podcast, bringing you all the latest news from Cloud and AI including AWS, Azure, Google, OCI, GPT-4.o, C7i-flex, AWS CEO Changes and Amazon Polly #aws #gcp #azure #cloud #cloudpodcast #podcast

18

This is a must see - 5mins of clear articulation of how and why Google Cloud differentiates in #GenAI with an unique approach and strategy relative to competitive #hyperscalers https://lnkd.in/gywMw7Dp

419

8 Kommentare

Back in 2000, I was an engineer, coding in C/C++ on Solaris. We used Clearcase for source control (yes, really). Builds and unit tests? All run locally before daring to touch the release branch. And merge conflicts? Don’t even get me started on those. 😂 Once everything was merged (fingers crossed), QA would spend a solid week testing before we could merge into the main branch. Every 3-6 months, we’d finally ship a release to our customers. The whole process was lengthy, painful, and wildly unproductive. Maybe that’s why I bailed on engineering! 😅 Fast forward to 2024, and the entire software development lifecycle fits into a day. How? DevOps, baby! And it all kicked off with Jenkins. Jenkins revolutionized the way we build and ship software. Now, we’re celebrating 20 years with the creator of Jenkins, Kohsuke Kawaguchi. Don’t miss the fireside chat with him and Sacha Labourey at DevOps World! If you haven’t registered yet, what are you even doing? Go sign up now (it takes a min): https://lnkd.in/eYMhG4ZM

48

4 Kommentare

Writing about the original IBM PC and the Intel 8080 processor reminded me of some fascinating news I came across last week. Microsoft has open-sourced MS-DOS 4.0. This version was originally developed in 1979, 45 years ago, in collaboration with IBM and was written entirely in 8086 assembly language. It is now available on Microsoft's GitHub repository. This release follows the earlier open-sourcing of MS-DOS 1.25 and 2.0, which Microsoft made available to the public 10 years ago. It’s remarkable that Microsoft has been around long enough to have its own 'Microsoft Archive.' I feel old 😉 For more details, check out the Microsoft blog post about the open-sourcing of MS-DOS 4.0: https://lnkd.in/g57iEA9x. If you're feeling adventurous, you can explore the MS-DOS open source repository here: https://lnkd.in/gj2aaCZA

1

I shamelessly want more people to read my post on Stateful Orchestration Engines! Here it is: https://lnkd.in/gVJ7tkj4 It probably goes against some unwritten social media rules to post about it here. But I published two blog posts last week, and am seeing disproportionately many reads on the first one compared to the second one. Could be because I shared them both in a single LinkedIn post. And it's this, second blog post which I believe is more important. So, lesson learned, will make N LinkedIn posts for N blog posts in the future. In this second blog post I introduce the notion of "stateful" functions, which may well be the next major step in how we design software. Last time this happened with `await`-ing on the results of `async` functions, which turned out to be a whole new paradigm of programming with green threads and the re-discovery of cooperative multithreading. There is even a new mainstream programming language whose popularity skyrocketed largely due to these await-friendly "coroutines" being a first-class citizen: Golang! And I'm a sucker for programming languages and DSLs. We do not have a perfect one, but there are plenty of good ideas, and these good ideas ought to converge in relatively good ways from time to time! If only we could get back to the good old "objects talking to each other via messages" paradigm, forgotten since the modern OOP replaced it. If only we could normalize thinking of Erlang in terms of async workflows, of Haskell in terms of the lack of side effects, and in terms of, well, Temporal-style ideas for long-running functions. So that developers would begin to truly reason about the code that is guaranteed to be executed no matter what, even if it `sleep`-s for thirty days in the meantime, and changes machines and execution runtimes several times while doing do. If only we could normalize exactly-once as the primary mode of thinking of complex operations in distributed systems, so that SAGA-s and compensating transactions and two-phase commits would be a thing of the past. I see quite some potential here. And am actively looking in this direction. And will appreciate it if you read this post, tell me what you think of it, and show it to friends and colleagues who are interested in shaping the future of tech in the same direction as I am.

17

📅 *Today* - Tuesday, October 22nd, 2024 11:00 AM PT [▶][▶] 👨‍💻👨‍💻Join Abhishek Narain and Wilson Lee for "Unlocking AI at Scale with Data Factory in Microsoft Fabric"! 📊 Efficiently managing high-volume, complex data is critical for building scalable AI solutions. Data Factory in Microsoft Fabric streamlines the process of ingesting, transforming, and orchestrating data, making it easier to prepare and integrate diverse data sources for AI applications 🔬Join this session to explore how Data Factory can help automate data workflows, enabling seamless data preparation for AI models. Learn how to design scalable data pipelines, automate data transformation, and optimize data integration, all within the Microsoft Fabric ecosystem. This is your chance to unlock the full potential of AI using Data Factory!. 🚀🔥Don’t miss this opportunity to learn and do more with AI in Microsoft Fabric. Register now for the Microsoft Fabric & AI Learning Hackathon and livestream series! Check out the "Microsoft Fabric & AI Learning Hackathon" on Devpost: https://lnkd.in/gwSAyWhE See the full "Microsoft Fabric & AI Learning Hackathon" livestream schedule on the MS Reactor: https://aka.ms/fabrichack Watch today's livestream event on-demand on YouTube: https://lnkd.in/gbN9qFTr

15

The complexities of the AI supply chain are a trade off on cost, performance and availability. The Hyperscaler partnerships have been a key component in AI distribution in the enterprise. But with new partnerships and new ecosystems... watch this space... https://lnkd.in/dtYyzESN

33

“A.L.I.C.E, can you get me this month’s publications on arxiv.org that talk about the latest pricing models and create a summary for each one of them?” “Okay, now can you give me an implementation for the first two in the list in Python?” #fintech #investmentmanagement #googlecloud

34

6 Kommentare

Platform-as-a-service (PaaS) solutions in the cloud are critical for the success of RAG and running AI models. Uwe Fassnacht explains why 👉 https://ibm.biz/BdKDVi

102

2 Kommentare

💡 Overcoming #interoperability challenges in hybrid and multi-clouds   A big shoutout to my DE-CIX colleague Dr. Christoph Dietzel for sharing his insights on the complexities cloud architects face when integrating disparate cloud environments. Many companies are dealing with isolated #cloudsolutions from different providers, leading to unexpected extra costs and interoperability headaches. Instead of reverting to a single-cloud policy, which can result in vendor lock-in and a single point of failure, he suggests a more robust #multicloud strategy.   The key takeaway? Direct #connectivity between clouds offers a secure, cost-efficient solution that enhances performance and reduces latency of data flows. Achieving interoperability on the network layer is essential for a resilient and flexible cloud environment. 🌐   🔍 How are you addressing interoperability challenges in your multi-cloud strategy?   https://lnkd.in/erETYsbU

36

Oracle today announced the general availability of HeatWave #GenAI, including in-database LLMs, an automated in-database vector store, and the ability to have human-like contextual conversations in natural language. https://lnkd.in/gK-Q44gc

3

AI-generated code can behave in unexpected ways if developers don't thoroughly review it! Dynatrace CTO Bernd Greifender explains six key reasons why AI-generated code could make it more difficult to maintain the reliability and security of software over the long term

11

Haters say that LLMs are useless. But Perplexity is now my go-to place for pretty much every non-trivial query I have. The ability to check the sources in a seamless fashion is simply brilliant. I only use Google occasionally and ChatGPT for summarization, brainstorming, etc. Proper UX and system design has always been and always will be the key to unlocking the potential of AI/ML. #ux #systemdesign #ai #ml #perplexity

23

2 Kommentare

It's fascinating to see how generative AI drives significant value in the competitive field of industrial operations. The post "Empowering predictive maintenance with Amazon Bedrock" by my colleagues Roberto Catalano and Luca Perrozzi demonstrates how AWS can help make things easier. #aws #iot #genai #amazonbedrock #industrial #manufacturing

4