Articles from Source: The-New-Stack

Web Development in 2025: AI’s React Bias vs. Native Web

2025-12-15 13:00
Web development is evolving in 2025, with a notable shift towards simplicity. Frameworks like Astro and Svelte are gaining traction as developers seek alternatives to React. Native web features are also advancing, matching JavaScript frameworks in functionality. The View Transition API is now widely available, enhancing web applications. AI-assisted coding is a major trend but often defaults to React due to its prevalence in training data. Stay updated with tools from the WebDX Community...
Source: The New Stack
Richard MacManus

The Real Bottleneck in Enterprise AI Isn’t the Model, It’s Context

2025-12-14 18:00
AI agents for data work are rapidly emerging, but many deployments are struggling due to a lack of understanding of the underlying data platform. These agents often query outdated or incorrect data, leading to inaccurate insights. This issue, termed the "context wall," highlights the gap between AI capabilities and operational reality. To improve effectiveness, a focus on orchestration is necessary. This creates a comprehensive operational record, enhancing agents' understanding of data...
Source: The New Stack
Julian LaNeve

Fedora Silverblue Has a Handy Tool To Help Simplify Development

2025-12-14 15:00
Explore Fedora Silverblue, a Linux distribution designed for developers! 🐧 It features Toolbox, a tool that creates containerized environments for software development and debugging. This keeps your host OS clean and stable while allowing easy access to different distributions. Toolbox is preinstalled, making it simple to get started. However, it's best for command line tools, not GUI apps. #FedoraSilverblue #Linux #Development #Toolbox #OpenSource
Source: The New Stack
Jack Wallen

Adafruit: Arduino’s Rules Are ‘Incompatible With Open Source’

2025-12-14 14:00
The open-source hardware community is in discussion over Arduino's new Terms and Conditions, following its acquisition by Qualcomm. 🤔 Adafruit has raised concerns that these changes restrict reverse engineering and impose broad monitoring on AI features. They emphasize that this threatens open-source principles. In response, Arduino clarified that restrictions only apply to their SaaS applications and affirmed their commitment to open-source hardware remains unchanged. The debate highlights...
Source: The New Stack
David Cassel

Three Core Principles for Sustainable Platform Design

2025-12-13 18:00
In the article "Three Core Principles for Sustainable Platform Design," key insights on effective platform engineering are discussed. A successful platform should provide software teams with contextualized tools, enabling them to focus on revenue-generating tasks. This requires a shift in how organizations deliver value, moving beyond just technology components. Three essential outcomes are highlighted for assessing platform value: 1️⃣ Quick access to needed resources. 2️⃣ Efficient...
Source: The New Stack
Abby Bangser

Rust-First Shuttle Launches Neptune for Easier App Deployments

2025-12-13 14:00
🚀 Exciting news in the tech world! Shuttle has launched Neptune, a Rust-first open-source platform designed to streamline app deployments. Neptune acts as an AI platform engineer, integrating with various coding tools and supporting major cloud providers like AWS, GCP, and Azure. It simplifies the deployment process by automating infrastructure provisioning while maintaining up-to-date specifications with your code. This beta version is now open to early builders. Meanwhile, security...
Source: The New Stack
Loraine Lawson

Linux Foundation Leader: We’re Not in an AI Bubble

2025-12-08 22:00
At the Open Source Summit Japan, Jim Zemlin, Executive Director of the Linux Foundation, discussed the state of AI investment. 💡 He highlighted that while AI may not be in a bubble, large language models could be, with projected spending on AI data centers reaching $3 trillion by 2028. 🌍 Zemlin emphasized the significant energy demands of AI, noting a substantial increase in Google's inference volume. He pointed out that the AI boom relies heavily on infrastructure, not just software. 🔌 Open-...
Source: The New Stack
Steven J. Vaughan-Nichols

The Automation Paradox and Why Security Teams Fear Their Own Tools

2025-12-08 21:00
Security automation is facing a challenge: speed has overshadowed precision. Many security teams hesitate to deploy automated responses due to fears of unintended consequences. The article discusses the "automation paradox," where tools that promise quick action often remain unused because of the risk of collateral damage. Unlike DevOps, which improved safety in automation, security operations still lack effective rollback plans. A proposed solution is "surgical containment," a framework that...
Source: The New Stack
Dana Rozen

Kubernetes Gets an AI Conformance Program — and VMware Is Already On Board

2025-12-08 20:00
🚀 The Cloud Native Computing Foundation has launched the Certified Kubernetes AI Conformance Program to tackle fragmentation in AI workloads. VMware’s vSphere Kubernetes Service is among the first to achieve certification, enhancing interoperability across platforms. Experts emphasize the importance of portability for seamless application movement between public and private clouds. Learn more about this initiative from KubeCon + CloudNativeCon! 🌐 #Kubernetes #AI #CloudComputing #VMware #CNCF
Source: The New Stack
Michelle Gienow

A Guide To Building Scalable AI Agents

2025-12-08 19:00
Discover how to build scalable AI agents with real-time decision-making systems! 🤖 This guide covers the integration of Salesforce Agentforce and Heroku AppLink. Agentforce uses AI to automate tasks while ensuring data security within your Salesforce environment. 📊 Learn to customize workflows and enhance automation with Heroku's AI PaaS for tailored applications. Improve efficiency and data access in your organization. 🌐 #AI #Salesforce #Automation #DataSecurity #Heroku
Source: The New Stack
Doug Sillars

How Generative AI Exposes a Weak Product Value Proposition

2025-12-08 18:00
Generative AI is reshaping user expectations, especially in creative industries like gaming. 🎮 Many products risk having their value questioned if they don’t clearly communicate their value proposition. Users are increasingly discerning about the quality of AI-generated content compared to human-created assets. As seen in games like ARC Raiders, players expect transparency about AI's role. If expectations aren't managed, it can lead to dissatisfaction. #GenerativeAI #GamingIndustry...
Source: The New Stack
David Eastman

Why the Model Context Protocol Won

2025-12-07 18:00
Anthropic's Model Context Protocol (MCP) launched in November 2024 as a universal standard for AI-to-tool integration, initially met with skepticism. In its early days, MCP mainly aided developers in improving AI-assisted coding but lacked broader connectivity. The protocol faced challenges with fragile transports and issues with authentication and timeouts. Despite existing AI-to-machine protocols, MCP gained traction as the community began to innovate and adapt it. #AI #Technology...
Source: The New Stack
Michael Yaroshefsky

AWS Tackles AI’s ‘Too Much Information’ Problem

2025-12-07 17:00
🚀 AWS introduces Kiro to address the issue of 'too much information' in AI development. With the new powers system, users can provide specific knowledge to agents without overwhelming them. This system activates tools based on current tasks, reducing unnecessary context loading. Developers can now focus on what they need, improving efficiency in coding. #AWS #AI #Kiro #Development #TechNews
Source: The New Stack
Darryl K. Taft

Reasons To Love Linux Mint

2025-12-07 15:00
Discover why Linux Mint is a favorite among new users! 🌟 This open-source operating system, based on Ubuntu, features the user-friendly Cinnamon desktop, resembling Windows for easy navigation. With preinstalled apps like Firefox and LibreOffice, users can start working or enjoying content right away. Linux Mint also offers helpful tools, such as the System Reports app, which simplifies tasks like installing multimedia codecs. Explore the benefits of Linux Mint today! 💻✨ #LinuxMint...
Source: The New Stack
Jack Wallen

Ignore Prior Instructions: AI Still Befuddled by Basic Reasoning

2025-12-07 14:00
AI language models (LLMs) face challenges in basic reasoning, according to Mark Russinovich, CTO of Microsoft Azure. He highlighted that LLMs often struggle with logical consistency and factual recall. 🧠 Russinovich emphasized the importance of understanding LLM limitations and the need for safety measures. While LLMs excel at summarization, they may overlook details and produce inconsistent results. ⚠️ As AI models evolve, improvements in reasoning ability are not guaranteed, raising...
Source: The New Stack
Joab Jackson

A Guide to Token-Efficient Data Prep for LLM Workloads

2025-12-06 19:00
🚀 As organizations adopt Retrieval-Augmented Generation (RAG) and AI systems, poor data serialization is causing significant token waste, consuming 40% to 70% of available tokens. This inefficiency leads to inflated API costs and reduced model performance, especially at scale. A single inefficient record can waste hundreds of tokens, impacting the viability of AI deployments. 💡 Understanding token optimization is essential. Key strategies include: 1. **Eliminating Structural Redundancy**:...
Source: The New Stack
Minav Suresh Patel

React Server Components Vulnerability Found

2025-12-06 15:00
🚨 A security vulnerability in React Server Components has been identified. Reported by Lachlan Davidson, this flaw allows unauthenticated remote code execution. The issue affects versions 19.0, 19.1.0, 19.1.1, and 19.2.0. Immediate upgrades to versions 19.0.1, 19.1.2, and 19.2.1 are necessary. If your app does not utilize a server or React Server Components, it is not impacted. #React #Security #Coding #DeveloperNews
Source: The New Stack
Loraine Lawson

Iterate.ai Launches AgentOne for Enterprise AI Code Security

2025-12-05 23:00
🚀 Iterate.ai has launched AgentOne, an autonomous coding assistant designed to enhance AI code security. This tool integrates security validation directly into the AI development process, addressing the challenge of rapid code generation that outpaces traditional security reviews. AgentOne operates within Visual Studio Code and Iterplay, featuring a Swarm Intelligence Architecture that conducts simultaneous security checks throughout development. This innovation aims to ensure secure coding...
Source: The New Stack
Darryl K. Taft

How Capital One Cut Tracing Data by 70% With OpenTelemetry

2025-12-05 22:00
Capital One recently showcased how they reduced tracing data by 70% using OpenTelemetry. During the Observability Day at KubeCon, engineers Joseph Knight and Sateesh Mamidala discussed the challenges of managing telemetry data. They emphasized the importance of optimal sampling to avoid data overload while ensuring accurate monitoring. Their approach involved implementing dedicated infrastructure for effective trace sampling, which streamlined operations across Capital One globally....
Source: The New Stack
B. Cameron Gain

Trae IDE Auto-Installs Python Libraries as You Code

2025-12-05 21:00
🚀 Discover the unique features of Trae IDE, which auto-installs Python libraries as you code! This IDE stands out by simplifying the coding process, especially for creating applications like a Dungeons & Dragons character sheet. Once installed, it prompts users to install necessary libraries automatically, saving time and effort. Trae is available for free on macOS and Windows, with a Linux version on the way. #TraeIDE #PythonCoding #AItools #DeveloperLife #DungeonsAndDragons
Source: The New Stack
Jack Wallen

How etcd Solved Its Knowledge Drain With Deterministic Testing

2025-12-05 20:00
The article discusses how etcd addressed knowledge loss when maintainers left the project. Lead maintainer Marek Siarkowicz highlighted that turnover led to reliability issues and critical bugs in new releases. To tackle this, the team implemented "robustness testing" and developed a framework inspired by Jepson. They aimed for linearizability in distributed systems, a significant challenge. Collaborating with Antithesis on deterministic simulation testing helped them better manage bugs and...
Source: The New Stack
Heather Joslyn

The Platform PM: Building an Ecosystem, Not Just a Product

2025-12-05 19:00
🚀 Platforms are more than just products; they create ecosystems for entire industries. They focus on integration rates, ecosystem health, and developer experience rather than solely on user growth. Platform PMs balance internal needs and external developer expectations, shaping environments for co-creation. Success is measured by integration speed and system reliability, much like city planning rather than app design. #PlatformManagement #EcosystemBuilding #ProductDevelopment #TechInnovation...
Source: The New Stack
Kateryna Korotieieva

JetBrains CEO on How Developers Become Leaders

2025-12-05 18:00
🚀 JetBrains CEO Kirill Skrygan shared insights on his journey from junior developer to CEO. Starting in tennis, he shifted to programming at a young age. He emphasizes the importance of impactful leadership and the role of mentorship in tech. Skrygan advocates for investing in juniors and interns to enhance team spirit and innovation. For those looking to transition from coding to management, his experience highlights the value of passion and dedication in the workplace. #Leadership...
Source: The New Stack
Loraine Lawson

AI Can Deliver Deployment-Aware Risk Analysis for Kubernetes

2025-12-05 17:00
🔍 Kubernetes platform engineers face a challenge with overwhelming alerts on security dashboards, often displaying thousands of critical vulnerabilities. Traditional risk scoring methods fall short by evaluating risks in isolation without considering the specific deployment context. Red Hat and IBM Research are addressing this by developing an AI-driven Risk Investigation Agent for more accurate, “deployment-aware” risk analysis. This innovation aims to refine risk assessments based on real-...
Source: The New Stack
Yair Allouche

AI Agents Are Morphing Into the ‘Enterprise Operating System’

2025-12-05 16:02
AI agents are evolving into essential tools within companies, acting as internal "operating systems" that streamline data flows across various software. At Block, the AI agent framework called goose connects with multiple tools like Slack and Google Drive, enhancing productivity for around 6,000 employees. This technology enables faster insights and decision-making, reducing reliance on traditional workflows. The focus is on empowering teams to solve problems independently, rather than...
Source: The New Stack
Angie Jones

DevOps Is Still Waiting for Its Cursor Moment

2025-12-05 15:00
🚨 Production alerts can disrupt your night. A recent article highlights the challenges engineers face when responding to incidents, such as service errors and system failures. While AI tools like Cursor have streamlined code writing, operations remain manual and complex. Engineers still juggle multiple tools and rely on tribal knowledge to resolve issues. The gap between development and operations continues to widen, with unique environments and mandatory governance adding to the complexity....
Source: The New Stack
Zafar Abbas

With Graviton5, AWS Promises a 25% Performance Boost

2025-12-04 20:00
🚀 AWS has unveiled its latest Graviton5 chips at the re:Invent conference, promising up to 25% better performance than the previous Graviton4. 🖥️ With 192 cores per chip and enhanced L3 cache, these chips aim to boost scalability and reduce latency for real-world applications. 🔒 The new Nitro Isolation Engine further improves workload isolation, enhancing security across AWS services. #AWS #Graviton5 #CloudComputing #TechNews #Innovation
Source: The New Stack
Frederic Lardinois

KubeCon Survey: How Platform Teams Are Adopting AI and IDPs

2025-12-04 19:00
🚀 KubeCon surveyed 219 platform engineers to explore the role of platform engineering in software organizations. Developers remain the primary focus, with 44.1% of respondents identifying them as key customers. Other stakeholders include SRE/Infra teams, data teams, and security teams. AI adoption is on the rise, with 30.1% of teams exploring AI code assistants. Future priorities include AI code reviews and automated ticket resolution. #KubeCon #PlatformEngineering #AI #SoftwareDevelopment...
Source: The New Stack
John Crowley

Stop Blaming React for Your State Management Hangover

2025-12-04 18:00
🚫 Stop blaming React for state management issues! Many developers face challenges not because of React itself, but due to misunderstandings in data flow. Relying on new state management libraries won't solve the core problem of unclear architecture. Overusing context can lead to complexity, making debugging difficult and impacting performance. It's essential to grasp unidirectional data flow to build effective and maintainable applications. #React #StateManagement #WebDevelopment #Coding...
Source: The New Stack
Alexander T. Williams

KubeVirt’s Architecture: CRDs, Controllers and Daemons

2025-12-04 16:00
🚀 KubeVirt is reshaping how organizations migrate to Kubernetes by bridging legacy virtual machines and modern containers. This architecture extends Kubernetes' capabilities, allowing VMs to coexist with containers on a single platform. KubeVirt introduces Custom Resource Definitions (CRDs) to manage VMs as native resources, simplifying orchestration. The recent eBook, “Running Virtual Machines on Kubernetes,” offers a comprehensive guide on this migration process. 📖 #KubeVirt #Kubernetes...
Source: The New Stack
Janakiram MSV

Your CI/CD Pipeline Is Not Ready To Ship AI Agents

2025-12-04 15:00
🚀 The article “Your CI/CD Pipeline Is Not Ready To Ship AI Agents” highlights a critical gap in AI infrastructure. While many companies are integrating AI features, true autonomous agents are still rare. Current delivery pipelines struggle with the complexities of non-deterministic software, making reliability a challenge. The shift from "prompt and pray" to deterministic workflows emphasizes the need for robust systems engineering and proper integration testing. #AI #CICD...
Source: The New Stack
Arjun Iyer

Mooncake Brings Databricks Rich Transactional Processing

2025-12-04 14:00
Databricks recently acquired Mooncake, enhancing its Lakebase platform for AI agents. 🌐 Mooncake offers rich transactional processing and fast analysis without managing ETL pipelines. This integration allows direct data access from PostgreSQL for decision-making. Developed by ex-SingleStore engineers, Mooncake aims to streamline the traditionally separate OLTP and OLAP systems, reducing manual data syncing efforts. #Databricks #AI #Mooncake #DataAnalytics #Lakebase
Source: The New Stack
Joab Jackson

What Adobe’s New AI Assistant Can Teach Frontend Developers

2025-12-03 23:00
Adobe's new AI Assistant aims to enhance accessibility in web and application development, according to Navya Agarwal, a senior software engineer at Adobe. She emphasizes that accessibility is not just altruism but a pathway to market growth. The AI Assistant, part of Adobe Express, integrates accessibility features from the start, addressing needs for users with cognitive disabilities, such as ADHD and dyslexia. It offers tailored design adjustments to improve user experience. This...
Source: The New Stack
Loraine Lawson

Welcome to AI’s Messy Middle: Where 36x Gains Require Distinguished Engineers

2025-12-03 22:00
Amazon Web Services CEO Matt Garman recently discussed Kiro, a new IDE, at AWS re:Invent. He highlighted a project that was completed in just 76 days by a small team, showcasing the rapid advancements in AI infrastructure. We are currently in the "messy middle" of AI development, where innovations are happening quickly but without established best practices. This presents challenges for companies as they adapt to new architectures and technologies. Garman also announced the availability of...
Source: The New Stack
Alex Williams

Combining Rust and Python for High-Performance AI Systems

2025-12-03 21:00
🚀 Python dominates AI and ML with its robust ecosystem, but it faces speed limitations due to its interpreted nature and global interpreter lock. 🌟 Rust offers high performance and memory safety, making it ideal for critical components in AI systems. By combining both, developers can prototype in Python while leveraging Rust for performance-intensive tasks. Learn how this hybrid approach is reshaping AI development and enhancing efficiency! #AI #MachineLearning #Rust #Python #HighPerformance
Source: The New Stack
Zziwa Raymond Ian

AWS Makes It Easier To Customize AI Models in Bedrock and SageMaker Without a Ph.D.

2025-12-03 19:30
🚀 AWS has introduced Nova Forge, simplifying the customization of its Nova large language models for enterprises. With new options in Amazon Bedrock and SageMaker, developers can now tailor AI models without needing advanced machine learning expertise. SageMaker offers a guided experience for customization, while Bedrock focuses on Reinforcement Fine Tuning for better task performance. No Ph.D. required! 🌟 #AWS #AI #MachineLearning #SageMaker #NovaForge
Source: The New Stack
Frederic Lardinois

How To Build AI Agents 3 Ways With Google ADK

2025-12-03 19:00
🚀 Google’s Agent Development Kit (ADK) is paving the way for AI agent creation! ADK, launched at Google Cloud NEXT 2025, supports agents in various Google products. Developers appreciate its flexibility, allowing for agent building through Python, YAML, or a visual interface. This tutorial guides you to create a “Hello World” agent using all three methods. Explore the distinct approaches: - **Imperative agents (Python)** offer flexibility and control. - **Declarative agents (YAML)** simplify...
Source: The New Stack
Janakiram MSV

When To Log, and When To Shut Up

2025-12-03 17:00
Many logs in software development can generate unnecessary noise. 🚫 The article emphasizes the importance of selective logging. Instead of logging every function or variable, developers should focus on entries that aid in debugging and understanding system performance. 📉 Context is key—logs should include relevant details like request IDs to be genuinely useful. By linking logs to traces and metrics, they become valuable evidence rather than clutter. 🔍 #Logging #Observability...
Source: The New Stack
Mike Shi

Helm 4: What’s New in the Open Source Kubernetes Package Manager?

2025-12-03 16:00
🚀 Helm 4 has officially launched at KubeCon + CloudNativeCon North America, marking the first major update in six years! Originally created as a hackathon project called Kate's Place, Helm has evolved into a vital open-source package manager for Kubernetes. The new version brings improved logging, dependency management, and WebAssembly plugins for enhanced portability. Learn more about its journey and features! 🌐 #Kubernetes #Helm4 #OpenSource #KubeCon #CloudNative
Source: The New Stack
Heather Joslyn

What DocumentDB Means for Open Source

2025-12-03 14:00
🚀 The open source community is focusing on DocumentDB for three main reasons. Firstly, it merges MongoDB's document features with PostgreSQL's relational capabilities, enhancing database flexibility. Secondly, DocumentDB is free under the MIT license and built on Microsoft Azure Cosmos DB's core. Lastly, it supports vital use cases for generative models and intelligent agents, offering schema-free advantages. #DocumentDB #OpenSource #DatabaseTechnology #MongoDB #PostgreSQL
Source: The New Stack
Jelani Harper