Articles from Source: The-New-Stack

The 2 failures with AI coding that are creating security bottlenecks

2026-03-11 16:15
AI coding tools have not met expectations, as a recent GitLab survey reveals. While over one-third of code is AI-generated, quality control and security vulnerabilities are major concerns for developers. ⚠️ As AI contributes to larger codebases, security teams face overwhelming review demands, creating bottlenecks. 🚧 Attackers are exploiting these vulnerabilities faster than teams can respond. To address these challenges, organizations must rethink their approach. It's important to integrate...
Source: The New Stack
Julie Davila

Publish your data, AI techniques, and agentic engineering work on Towards Data Science

2026-03-11 15:26
🚀 Have you developed an agent pipeline or improved AI model reliability? Consider sharing your expertise on Towards Data Science! This platform invites practitioners to publish their work, offering a chance to reach a large audience of data scientists and AI engineers. Submissions are free, and authors can earn through the TDS Author Payment Program. Join the community and contribute your insights! #DataScience #AI #MachineLearning #TechWriting #TDS
Source: The New Stack
Ludovic Benistant

Amazon calls engineers for a “deep dive” internal meeting to discuss “GenAI”-related outages

2026-03-10 20:09
Amazon recently held an internal meeting to address a series of AI-related outages impacting its website and app. 📉 The company's top retail executive called engineers together to investigate these incidents, which have reportedly been linked to AI-assisted coding errors. Over the past week, customers faced significant downtime, unable to check out or access account information. 🛒 Amazon's SVP of eCommerce noted that recent GenAI changes have led to unsafe practices, prompting a review of...
Source: The New Stack
Meredith Shubel

With its latest Phi-4 reasoning model, Microsoft reckons bigger isn’t always better

2026-03-10 19:51
Microsoft's new Phi-4-Reasoning-Vision-15B model challenges the trend of larger AI systems. By focusing on compact models trained on high-quality reasoning data, Microsoft aims to demonstrate that size isn't the only measure of capability. This multimodal model can process both text and images effectively. Released with open weights, developers can access Phi-4 on platforms like Hugging Face and GitHub, promoting further experimentation. #AI #MachineLearning #Microsoft #Innovation #PhiModels
Source: The New Stack
Paul Sawers

Nvidia plans NemoClaw launch, an open-source platform for AI agents

2026-03-10 18:35
🚀 Nvidia is set to launch NemoClaw, an open-source platform for AI agents, next week. This initiative aims to assist companies in deploying AI agents to enhance employee productivity. 🤖 Major companies like Salesforce and Google are already being approached to utilize this platform, which is accessible to all, regardless of hardware. 🔒 With added security measures, Nvidia seeks to address concerns regarding AI agent safety. The unveiling will occur at the GTC developer conference in San Jose...
Source: The New Stack
Meredith Shubel

How to deploy an AI server on your Debian/Ubuntu server

2026-03-10 18:00
🚀 Setting up an AI server on your Debian/Ubuntu is easier than you think! Locally installed AI offers two main benefits: reduced power demand from the grid and enhanced privacy for your data. To get started, ensure you have a running Debian or Ubuntu Server and a user with sudo privileges. Installation steps include adding your user to the Docker group, installing Ollama, and configuring it for remote connections. Ready to dive into the setup? Check out the full guide! #AI #Debian #Ubuntu...
Source: The New Stack
Jack Wallen

With GridGain acquisition, MariaDB bets on in-memory computing and Apache Ignite

2026-03-10 13:47
MariaDB has announced its acquisition of GridGain Systems to enhance in-memory computing capabilities. This move aims to address the growing demands of AI applications, which require faster data access. Vikas Mathur, MariaDB's CPO, highlighted the challenge of AI latency due to fragmented data across systems. The integration of MariaDB’s database with GridGain’s technology seeks to empower organizations to deploy AI applications more effectively. 🌐💡 #MariaDB #GridGain #AI #InMemoryComputing...
Source: The New Stack
Paul Sawers

The AI Infrastructure crisis: When ambition meets ancient systems

2026-03-10 11:00
Artificial Intelligence is reshaping modern software, yet many organizations struggle with outdated legacy systems. 🏢 These systems, designed for traditional tasks, cannot meet AI's demands for performance and scalability. This leads to prolonged project timelines and missed opportunities. ⏳ To succeed, companies must shift their mindset around data and infrastructure, recognizing data as a strategic asset and embracing flexible ecosystems. 🌐 Building AI-ready infrastructure involves four...
Source: The New Stack
Zziwa Raymond Ian

Anthropic launches a multi-agent code review tool for Claude Code

2026-03-09 19:00
🚀 Anthropic has launched Code Review for Claude Code, a multi-agent tool designed to enhance code reviews. This new feature helps developers identify bugs before human review, addressing the growing demand for code oversight. It operates in the cloud for enabled repositories, providing insights with a focus on logical errors to minimize false positives. Since implementing it internally, Anthropic has seen a significant increase in substantive review comments on pull requests. #CodeReview...
Source: The New Stack
Frederic Lardinois

How context rot drags down AI and LLM results for enterprises, and how to fix it

2026-03-09 16:00
Data is essential for business success, but context is increasingly crucial for AI performance. The issue of "context rot" arises when new data conflicts with older information, causing confusion in AI and LLMs. This overload can lead to delays and incorrect answers as AI struggles to find relevant information. Understanding and managing this context can enhance AI effectiveness in enterprises. #AI #DataManagement #ContextRot #BusinessIntelligence #LLM
Source: The New Stack
Todd R. Weiss

Cursor builds always-on agents to tackle developer task tedium

2026-03-09 15:05
🚀 Cursor has launched Cursor Automations, introducing always-on agents designed to simplify repetitive developer tasks. These agents leverage deep codebase understanding, enhancing features like Plan Mode and predictive autocomplete. They can automate processes such as incident triage and code cleanup. Agents can be triggered by events like Slack messages or GitHub pull requests, streamlining workflow. For more insights, check out the details from Cursor's team! #Cursor #AI #Automation...
Source: The New Stack
Adrian Bridgwater

Moving AI apps from prototype to production requires enterprise-grade postgres infrastructure

2026-03-09 14:00
The rise of AI is undeniable, with 78% of organizations using it in 2024, a significant increase from the previous year. However, many struggle to measure ROI, as 90% of tech leaders admit challenges in evaluating their investments. 📊 Transitioning from AI prototypes to production poses additional difficulties. While new coding tools make development easier, they don’t guarantee that applications will meet the demands of enterprise environments, especially in regulated sectors like finance...
Source: The New Stack
Meredith Shubel

AI coding agents can write code, Crafting wants to help them ship it

2026-03-09 14:00
AI coding agents are advancing in generating code, but challenges remain in testing and deployment. Crafting, a San Francisco startup, aims to address this by providing AI agents with production-like environments for better code validation. Their platform allows agents to operate with real dependencies and data. Recently, Crafting announced the launch of Crafting for Agents, enhancing their existing cloud-based tools for engineers. This aims to streamline the engineering process for large...
Source: The New Stack
Frederic Lardinois

The technical leap where most brilliant AI initiatives spectacularly fail

2026-03-09 11:00
Navigating AI implementation is challenging. While many organizations overcome legacy infrastructure, the leap to production remains difficult. Did you know that 85% of AI projects never reach production? Many models fail under real-world conditions despite initial success in testing environments. AI initiatives require robust systems engineering to handle massive data, serve predictions swiftly, and ensure reliability. Understanding the unique scaling demands of AI is critical for success....
Source: The New Stack
Zziwa Raymond Ian

Vibe code full-stack apps fast with TanStack Start

2026-03-08 17:00
Unlock the potential of full-stack app development with TanStack Start! 🚀 This framework simplifies the process with type-safe routing, server functions, and file-based routing, making it beginner-friendly. It allows developers to create apps quickly without complex configurations. With features like Streaming SSR for faster rendering and secure server-side logic, TanStack Start empowers you to focus on creativity while coding. Ready to vibe code? 🌐💻 #TanStackStart #FullStackDevelopment...
Source: The New Stack
Jessica Wachtel

Vibe code full-stack apps fast with TanStack Start

2026-03-08 17:00
🚀 Building full-stack applications has never been easier! TanStack Start offers a streamlined approach for developers, especially beginners. It features type-safe routing, server functions, and file-based routing, allowing for quick app development without complex setups. Key aspects include: - Type-Safe by Default - Simplified File-Based Routing - Integrated Server Functions - Efficient Streaming SSR Explore how TanStack Start can enhance your development process! 🌐💻 #TanStack...
Source: The New Stack
Jessica Wachtel

Why is your Kubernetes cluster adding nodes when the dashboards look fine?

2026-03-08 15:10
Kubernetes clusters are increasingly adding nodes despite seemingly adequate resource utilization. This trend is linked to the rise of bursty workloads, especially in AI applications, which puts pressure on scheduling and autoscaling behaviors. Even with tools like Cluster Autoscaler or Karpenter, misconfigured inputs can lead to unexpected scaling. Metrics may appear calm, but if requests are set too high and not adjusted over time, pods can end up pending, prompting the cluster to scale...
Source: The New Stack
Yasmin Rajabi

Snowflake Cortex Code CLI adds dbt and Apache Airflow support for AI-powered data pipelines

2026-03-08 13:00
🚀 Exciting news from Snowflake! The Snowflake Cortex Code CLI has added support for dbt and Apache Airflow, enhancing its utility for AI-powered data pipelines. This move marks a significant step in extending capabilities beyond Snowflake's ecosystem. 🌐 By introducing a self-service subscription model, Snowflake is making the coding agent accessible to a broader audience, including non-customers. This aligns with their commitment to the open-source community. Agent Skills for debugging and...
Source: The New Stack
Jelani Harper

NanoClaw can stuff each AI agent into its own Docker container to deal with OpenClaw’s security mess

2026-03-07 18:00
🌐 NanoClaw addresses security concerns with OpenClaw by isolating each AI agent in its own Docker container. This approach ensures that agents operate independently, enhancing security measures. 📦 The application uses minimal code rather than large configuration files, allowing for efficient operations. Claude can adapt its code as needed, keeping the overall size manageable. 🔗 However, connecting to platforms like WhatsApp poses challenges due to strict policies. The preferred method for...
Source: The New Stack
David Eastman

Is AI Killing Open Source Software?

2026-03-07 17:00
AI's impact on open source software is complex. 🌐 Stormy Peters from AWS highlights that while AI isn't killing open source, it challenges long-held social and economic norms. AI-generated code is flooding maintainers with low-quality submissions, complicating their work. Interestingly, many developers are still contributing AI-generated code back to projects, despite concerns about its value. #OpenSource #AI #TechTrends #SoftwareDevelopment #Innovation
Source: The New Stack
Steven J. Vaughan-Nichols

Open-source coding agents like OpenCode, Cline, and Aider are solving a huge headache for developers

2026-03-07 14:00
AI coding agents are changing the landscape for developers, but the costs of running large language models (LLMs) are becoming challenging. Open-source solutions like OpenCode, Cline, and Aider help tackle these issues by providing a consistent interface that works across various LLMs. OpenCode recently introduced a $10/month subscription to simplify workload management. These tools interpret tasks, navigate codebases, and streamline model calls, marking a new trend in developer tooling. 🌐💻...
Source: The New Stack
Paul Sawers

OpenAI GPT-5.4 launches, AI gets its own jobs report, Claude surges after U.S. ban

2026-03-07 13:55
OpenAI has launched GPT-5.4, its latest model, boasting enhanced capabilities for professional tasks, with impressive benchmark scores. 📊 Alongside this, Codex has reached 1.6 million active users after its release on Windows, emphasizing improvements in developer workflows. 💻 A study by Anthropic reveals that AI is reshaping jobs, particularly for young workers, but overall unemployment remains stable. The focus is shifting towards systems design and strategy rather than job loss. 🔄 #AI...
Source: The New Stack
Matthew Burns

OutSystems CEO on how enterprises can successfully adopt vibe coding

2026-03-06 20:53
Enterprises are exploring AI agents, but OutSystems CEO Woodson Martin emphasizes that success involves more than just standalone agents. He highlights the need for a mix of agents, data, workflows, and human involvement to achieve effective outcomes. Some companies are already seeing positive results, like Travel Essence, which reduced customer planning time significantly. The use of AI in decision support and document processing is also gaining traction. 🔍🤖📈 #AIAgents #LowCode #Automation...
Source: The New Stack
Frederic Lardinois

Anthropic and OpenAI are battling for the best open-source maintainers

2026-03-06 20:19
OpenAI and Anthropic are competing to attract open-source maintainers, crucial in today's software landscape. OpenAI recently reopened applications for its Codex for Open Source program, offering benefits like six months of ChatGPT Pro and API credits for eligible projects. Anthropic's Claude for Open Source Program provides similar perks, including free access to its premium subscription for select maintainers. Both programs aim to integrate their tools into popular projects, enhancing...
Source: The New Stack
Paul Sawers

IT-Tools brings many useful developer tools into one convenient location

2026-03-06 20:00
🔧 Developers often use many tools for their work, from token generators to Docker converters. IT-Tools consolidates over 80 essential tools into one easy-to-access platform, streamlining your workflow. You can even create custom tools! To get started, you'll need a Docker-compatible server. The article provides detailed steps for installing Docker on Ubuntu Server 24.04. Check it out for a more efficient development experience! 🖥️🚀 #DeveloperTools #ITTools #Docker #Coding #TechEfficiency
Source: The New Stack
Jack Wallen

Nearly half of all companies now use Rust in production, survey finds

2026-03-06 18:45
🚀 The 2025 State of Rust Survey reveals that nearly 50% of organizations now use Rust in production, marking significant growth from 38.7% in 2023. 📊 Conducted from November to December 2025, the survey highlights a maturing community with 55.1% of developers using Rust daily. 🔍 Top deployment areas include server-side applications (51.7%) and cloud computing (25.3%). 💡 Interestingly, many non-users cite time constraints as the main reason for not adopting Rust. #Rust #Programming #TechTrends...
Source: The New Stack
Darryl K. Taft

Long-term support for Linux releases gets a new lease on life

2026-03-06 17:30
🔒 To keep Linux instances secure, Long Term Support (LTS) kernels are essential. 📅 In 2023, LTS support was reduced to two years due to maintainer burnout. However, user feedback led to a decision to extend support for key releases. 🔧 Linux 6.6 will now be supported until 2027, while 6.12 and 6.18 will last until 2028. This provides a longer support window compared to the previous two-year plan. 📉 Older 5.xx branches will retain their end-of-life dates in December 2026. #Linux #LTS...
Source: The New Stack
Steven J. Vaughan-Nichols

The case for running AI agents on Markdown files instead of MCP servers

2026-03-06 16:40
A venture capitalist is successfully operating his entire company using just twelve Markdown files, avoiding traditional web applications and workflow engines. 📝 These files encode essential workflows, managing tasks like drafting emails and handling support tickets. CompanyOS connects to various APIs, but the core intelligence relies on Markdown documents. Industry experts suggest many MCP servers may be unnecessary, as they often replicate functions that can be efficiently handled by skill...
Source: The New Stack
Janakiram MSV

One developer, team power: The future of AI-driven DevSecOps

2026-03-05 22:29
AI is changing the landscape of software delivery, but effective collaboration remains essential. While AI tools can enhance individual developer capabilities, they cannot replace the need for teamwork. Engineers must still evaluate AI-generated outputs, ensuring quality and security. The future of DevSecOps lies in fostering a collaborative culture that promotes knowledge-sharing and structured mentoring. This dual-gate system balances speed and risk, driving better software outcomes....
Source: The New Stack
Bryan Ross

OpenAI launches GPT-5.4 Thinking and Pro

2026-03-05 18:00
🚀 OpenAI has launched GPT-5.4, its latest and most capable model yet. This version integrates the coding strengths of GPT-5.3-Codex with enhanced support for documents and spreadsheets. The model improves web searching and context maintenance, with claims of reduced errors by 18% and false information by 33% compared to GPT-5.2. Notably, GPT-5.4 introduces a planning feature for real-time adjustments during responses, enhancing user control. Pricing sees an increase, with GPT-5.4 Pro being...
Source: The New Stack
Frederic Lardinois

Sam Altman wonders: Could the government nationalize artificial general intelligence?

2026-03-05 13:00
Sam Altman recently discussed the idea of government involvement in the development of artificial general intelligence (AGI). He suggested that it might be beneficial for AGI to become a government project, although he acknowledged that this outcome seems unlikely given the current trajectory. 🤔 During a Q&A session, he emphasized the importance of strong partnerships between AI companies and government entities. Altman noted the historical context of government-funded projects, like the...
Source: The New Stack
David Cassel

Why enterprise software development needs air traffic control

2026-03-04 22:35
Navigating the evolving landscape of AI tools in enterprise software development presents challenges for CIOs. With developers eager to utilize new AI coding assistants, the rapid pace of innovation complicates decision-making. Enterprises must balance agility with data privacy and compliance. Tool fragmentation is a significant issue, with many teams using multiple tools, leading to inefficiencies. 📉 The article emphasizes the need for effective management of these tools, likening it to air...
Source: The New Stack
Emilio Salvador

AerynOS is a Linux distribution geared toward performance and bulletproof updates

2026-03-04 20:00
Discover AerynOS, an independent Linux distribution designed for speed and reliable updates. 🚀 This OS features a next-gen package manager called Moss and a rapid build system named Boulder, streamlining app creation. It also simplifies complex EFI configurations to reduce boot issues. Currently in early development, AerynOS is not recommended for daily use, but is perfect for testing and feedback. It offers a choice of desktop environments, including GNOME and COSMIC. Try it out and explore...
Source: The New Stack
Jack Wallen

OpenAI’s Codex is now on Windows

2026-03-04 19:01
🚀 Exciting news! OpenAI’s Codex coding app is now available on Windows. Originally launched for Mac, it quickly gained popularity with over 1 million downloads in its first week. There’s strong anticipation for the Windows version, with over 500,000 developers on the waitlist. The Windows app offers native features like sandboxing and supports familiar tools. It maintains a similar look to the Mac version but includes Windows-specific skills such as WinUI for app developers. Codex uses...
Source: The New Stack
Frederic Lardinois

GSMA Open Gateway offers developers one API for 300+ mobile networks

2026-03-04 18:26
GSMA has launched the Open Gateway initiative, providing developers with a standardized API for over 300 mobile networks. 🌍📱 This approach allows software engineers to access telco capabilities like location and security without needing to adapt to different network protocols. With support from 85 operator groups, the initiative aims for practical implementation in mobile app development. 🔧 Learn more about how this can streamline connectivity for developers! #GSMA #OpenGateway...
Source: The New Stack
Adrian Bridgwater

The AI Shift: Why RISC-V is poised to challenge Arm and x86

2026-03-04 18:00
The RISC-V standard is evolving, reducing performance gaps with established Arm and x86 processors. This change supports cost-effective chip programming and offers developers freedom from licensing fees. RISC-V's vector capabilities make it ideal for AI applications, especially for parallel processing tasks. Notably, NVIDIA announced CUDA support for RISC-V, enhancing compatibility with Linux systems. This marks a significant step in open-source migration for RISC-V devices. #RISCV #AI...
Source: The New Stack
B. Cameron Gain

Why traditional ITOps is failing to keep up with the unique nature of AI incidents

2026-03-04 18:00
AI adoption is skyrocketing, with 88% of organizations using it in their operations. As we approach 2026, the focus shifts to managing AI incidents effectively. Organizations must treat AI incidents as a unique category, implementing specific processes for remediation to address new risks. Many lack governance policies, highlighting the need for responsible AI practices. Key metrics like hallucination rates and bias are now vital for assessing AI reliability. Cross-functional training will...
Source: The New Stack
Kat Gaines

DragonflyDB CEO: Most real-time AI infrastructure was built for a different era

2026-03-04 17:44
Real-time AI is crucial for personalization and revenue. However, many existing infrastructures struggle to meet modern demands. Legacy systems can lead to performance issues, operational complexity, and rising costs. A robust real-time data layer is essential for supporting intelligent systems effectively. Join DragonflyDB CEO Oded Poncz on March 10 for a live webinar on overcoming these challenges. 🔗 Register today! #RealTimeAI #MachineLearning #Webinar #DataInfrastructure #AIInsights
Source: The New Stack
TNS Staff

Cloud repatriation is hard. Here’s how to build a self-service developer platform that works.

2026-03-04 17:14
As organizations shift workloads back on-premises, they face challenges in balancing cost, security, and cloud-like speed. 🤔 To avoid bottlenecks and frustration among developers, a robust internal developer platform is crucial. It should be integrated and automated for ease of use while minimizing maintenance issues. Join the live webinar on March 12 at 10:30 a.m. PT to learn how VMware vSphere Kubernetes Service can streamline your development process. 🚀 #CloudRepatriation...
Source: The New Stack
TNS Staff

Eclipse Foundation reports Open VSX hits 300 million monthly downloads

2026-03-04 15:21
🚀 The Eclipse Foundation has announced that Open VSX, the open-source extension registry for VS Code, has reached 300 million monthly downloads. This milestone highlights its importance in the developer ecosystem, especially amid growing demand from AI and cloud-based platforms. Amazon Web Services (AWS) has committed a multi-year, seven-figure investment to support Open VSX and enhance its infrastructure. This investment is separate from AWS's membership in the foundation. #OpenSource...
Source: The New Stack
Paul Sawers