Articles from Source: The-New-Stack

How MIT’s Project NANDA Aims To Decentralize AI Agents

2025-10-24 15:00
🚀 MIT's Project NANDA is working to decentralize AI agents. In 2025, the focus has been on developing tools for these agents, but distribution remains a challenge. NANDA aims to create an "Internet of AI Agents" with cross-platform discovery and federated registries. The project builds on existing protocols and features four layers: discovery, identity, federation, and interoperability, to ensure autonomous and verifiable agents. #MIT #AI #Decentralization #ProjectNANDA #Innovation
Source: The New Stack
Richard MacManus

Beyond Basic Scaling: Advanced Kubernetes Resource Strategies

2025-10-24 14:30
Navigating Kubernetes resource management can be challenging. Overprovisioning wastes resources, while underprovisioning frustrates developers and slows down product delivery. ⚙️ The right balance is crucial for application stability and efficient cluster utilization. A reliable, automated resource management system can help teams optimize their Kubernetes environment. Join the free webinar on Oct. 21 at 11 a.m. PT to learn best practices and strategies for effective resource management. 📅...
Source: The New Stack
Vicki Walker

Multipass: Fast, Scriptable Ubuntu VMs for Modern DevOps

2025-10-24 14:00
Unlock the power of Ubuntu VMs with Canonical's Multipass! 🚀 This tool simplifies the launch and management of lightweight Ubuntu virtual machines across macOS, Windows, and Linux. It's designed for developers and ops teams, enabling quick VM provisioning through command line or scripts. Multipass features a client-server architecture for enhanced security and automation. It smartly selects the best hypervisor for your OS, optimizing performance and resource management. Ready to streamline...
Source: The New Stack
Janakiram MSV

When To Use Synthetic Monitoring vs. Real User Monitoring

2025-10-24 13:00
Understanding when to use synthetic monitoring versus real user monitoring (RUM) is essential for web performance. Most teams start with synthetic monitoring for predictable testing. However, real user feedback often reveals hidden issues like slow performance on mobile or specific browsers. Synthetic monitoring is great for launching new features, while RUM provides insights on actual user experiences. Both methods have their strengths, and knowing when to use each is key. #WebPerformance...
Source: The New Stack
Virna Sekuj

Securing the Vibe: Reducing the Risk of AI-Generated Code

2025-10-24 12:24
Vibe coding empowers developers to quickly turn ideas into applications. However, this speed can compromise security. AI-generated code often has vulnerabilities that require human review before deployment. To tackle security risks, organizations should apply threat models like STRIDE and utilize checklists like OWASP Top 10 for LLM Applications. By doing so, developers can harness the benefits of AI while maintaining security. ⚙️💻🔍 #AICoding #Cybersecurity #VibeCoding #SoftwareDevelopment #AI
Source: The New Stack
Crystal Morin

Unblock DevEx by Getting Observability Tools Out of Their Way

2025-10-23 22:00
🚀 Developers often face challenges with observability tools that disrupt their workflow. When these tools are complicated, it leads to lower productivity and increased frustration. Tools should simplify processes and integrate seamlessly into existing workflows to enhance developer experience. Join the free webinar on Nov. 6 to learn how to shift from friction to flow in observability. #DevEx #Observability #Webinar #DeveloperTools #Productivity
Source: The New Stack
Vicki Walker

Why Kubernetes Security Is Critical for GenAI Integrity

2025-10-23 21:00
As generative AI adoption grows, Kubernetes is becoming essential for managing these workloads. However, this shift brings significant security challenges. 🔒 A recent CNCF report indicates that 76% of organizations see security as their top concern, with unauthorized access and misconfigurations posing major risks. These vulnerabilities can lead to data leaks and intellectual property theft. Kubernetes' dynamic nature complicates security, making it hard to maintain visibility and control....
Source: The New Stack
Utpal Bhatt

AI Agents Will Eat Enterprise Software, Just Not in One Bite

2025-10-23 20:03
AI is transforming enterprise software, but it's not an overnight change. 🚀 While some believe AI agents will completely replace traditional solutions, this overlooks the complexities of software use in businesses. Currently, AI excels at simpler tasks and assists humans in more elaborate processes. For now, AI agents work well in research and defined tasks, but they require structured processes to ensure reliability. As technology advances, their role in handling complex tasks will evolve....
Source: The New Stack
Peter White

How OpenAI’s AgentKit Embeds ChatGPT Into Any Website

2025-10-23 19:03
🚀 OpenAI has introduced AgentKit, a toolkit for developers to integrate ChatGPT-powered agents into websites and applications. AgentKit features two main components: the Responses API, which manages outputs and tool integrations, and the Agents SDK, which oversees conversation states and orchestrates multi-step tasks. The architecture streamlines development, ensures reliability, and enhances security with built-in guardrails to protect sensitive data. Developers can customize these features...
Source: The New Stack
Janakiram MSV

How To Deploy a Local AI via Docker

2025-10-23 18:00
Unlock the potential of local AI with Docker! 🖥️ If you're concerned about data privacy and environmental impact, deploying AI locally is an ideal solution. The article outlines how to set up a local AI service using Docker for enhanced privacy and easier management. You'll need a compatible operating system (Linux, macOS, or Windows) and some basic commands to install Docker and deploy your AI efficiently. Explore methods for deployment with or without an NVIDIA GPU to tailor your setup to...
Source: The New Stack
Jack Wallen

Speed Plus Safety: Ending the Dev vs. Platform Engineer Clash

2025-10-23 17:00
In the cloud-native landscape, developers and platform engineers often face a cultural clash. Developers prioritize speed and innovation, while platform engineers focus on safety and stability. This difference can lead to friction and slowdowns in the development process. Key issues include access friction, delivery friction, and cognitive friction, all of which hinder efficiency and create bottlenecks. Both roles are crucial, but aligning their goals is essential for smoother operations. 🔧🚀...
Source: The New Stack
Jeroen van Erp

Lightning AI Brings a PyTorch Copilot to Its Development Environment

2025-10-23 16:30
🚀 Lightning AI has announced new tools and features to enhance its PyTorch development environment. The updates include an AI code editor in Lightning Studios and Notebooks, designed to assist developers with training and reinforcement learning tasks. Additionally, the new Lightning Environments provide scalable, sandboxed spaces for exploring AI workloads. This aims to simplify the development process for AI teams and expand access to PyTorch capabilities. #LightningAI #PyTorch...
Source: The New Stack
Frederic Lardinois

The Rise of JavaScript in Machine Learning

2025-10-23 16:03
JavaScript developers can leverage machine learning (ML) without needing to master Python. 🖥️ Laurie Lay from Ippon Technologies highlighted this at the recent devmio International JavaScript Conference. She explained that while Python dominates ML due to its libraries, JavaScript offers new opportunities for AI integration in applications. Python's success is linked to its role as a high-level glue language, enabling easier access to powerful libraries like NumPy and Pandas. These libraries...
Source: The New Stack
Loraine Lawson

Breaking the ‘Shared-Nothing’ Bottleneck: A NoSQL Paradigm

2025-10-23 15:00
Exploring the NoSQL paradigm reveals that a shared-nothing architecture is often favored for its high performance and low latency. 🖥️ Using direct-attached storage (DAS), NoSQL databases like Cassandra and MongoDB can achieve efficient data management. However, DAS can hinder sustainability efforts due to increased hardware needs and underutilization. 🌱 Modern SAN solutions provide a way to maintain performance while enhancing efficiency and resilience. 🔄 #NoSQL #DataManagement #TechTrends...
Source: The New Stack
Carol Platz

Are Unikernels the Answer for Next-Gen AI Cloud Workloads?

2025-10-23 14:00
🚀 Vercel has invested in Unikraft, a unikernel-based cloud-hosting company, to enhance app startup speeds and database performance. Prisma is already utilizing unikernels to run PostgreSQL instances efficiently, allowing them to offer a free tier for users. This technology enables low-latency applications and optimizes resource usage by scaling down unused instances. Unikernels differ from containers by operating as microVMs with their own minimal OS, rather than sharing the host OS....
Source: The New Stack
Joab Jackson

Rust vs. C++: a Modern Take on Performance and Safety

2025-10-22 21:00
🔍 Exploring the world of systems programming, this article compares Rust and C++. C++ has dominated for decades, offering developers extensive control over memory and performance. However, this power comes with risks, including potential crashes and security vulnerabilities. Rust, backed by Mozilla, aims to provide similar speed with enhanced memory safety through its unique ownership model. It enforces safety rules at compile time, reducing issues like data races and undefined behavior. The...
Source: The New Stack
Zziwa Raymond Ian

Why Up to 70% of Platform Engineering Teams Fail to Deliver Impact

2025-10-22 20:05
🚨 A recent article highlights the challenges faced by platform engineering teams, with 60-70% failing to deliver impact. Many teams struggle to align their goals with business objectives, leading to high rates of disbandment within 18 months. Insights from experts at Fast Flow Conf emphasize the importance of a product mindset and clear stakeholder alignment to achieve long-term success. #PlatformEngineering #TechChallenges #BusinessGoals #FastFlowConf #EngineeringInsights
Source: The New Stack
Jennifer Riggins

Why AI Is Generating ‘Lowest Common Denominator’ React Code

2025-10-22 19:00
AI is changing the landscape of React development, according to Seth Webster of the React Foundation. He argues we're moving toward a post-frontend-framework world, where AI-generated code lacks quality. Large language models are primarily trained on subpar React code, limiting their effectiveness. This results in AI acting like a mid-level engineer, using practices that may not align with best coding standards. Webster emphasizes the importance of structuring business logic separately from...
Source: The New Stack
Loraine Lawson

Paved Roads, Golden Paths, Guardrails and Railroads

2025-10-22 18:00
Paved roads, golden paths, guardrails, and railroads are essential components of developer workflows. These frameworks help streamline processes, reduce cognitive load, and enhance efficiency. 🚧💻 Paved roads represent standardized tools and practices, guiding developers through the software delivery life cycle. They promote autonomy and cooperation while ensuring consistency across teams. 🤝📊 However, challenges like perpetuating bad habits and transferring pain points can arise if not...
Source: The New Stack
Dario Esposito

AI Agents in IT: From Hype To Hands-on Impact

2025-10-22 17:00
Organizations are increasingly adopting AI to enhance productivity and efficiency. In 2024, investments reached $252.3 billion, yet most companies report modest financial benefits, with cost savings under 10% and revenue growth below 5%. To realize AI's full potential, enterprises must integrate AI throughout the software development life cycle. This includes incident remediation, predictive monitoring, and application performance. IT operations are ideal for AI agents, as these tools help...
Source: The New Stack
Bill Lobig

Ray Comes to the PyTorch Foundation

2025-10-22 16:00
🚀 Exciting news for the AI community! The PyTorch Foundation has announced that it will host Ray, a leading open-source distributed computing framework. Ray enhances AI and Python applications by providing tools for building scalable systems. Originally developed at UC Berkeley, Ray has been commercialized by Anyscale, which has raised significant funding to enhance its offerings. This addition strengthens the PyTorch Foundation’s portfolio, complementing existing projects like PyTorch and...
Source: The New Stack
Frederic Lardinois

Can the 50-Year-Old Actor Model Rescue Agentic AI?

2025-10-22 14:47
In 2025, agentic AI is expected to transition from research to production, with Gartner reporting that 34% of businesses already use AI agents. By 2028, 15% of daily work decisions could be made autonomously. However, many projects currently face challenges, often due to infrastructure issues rather than AI intelligence. The actor model, introduced in 1973, could provide a solution by enabling efficient management of autonomous processes. This model allows for lightweight, independent...
Source: The New Stack
Kimberley Mok

The Challenges of Uniting VMs and Containers on a Single Platform

2025-10-22 14:00
Kubernetes is recognized as a key platform for containerized applications, but there's a growing conversation about its role in managing virtual machines (VMs) as well. This shift raises important issues around skills and migration strategies. While combining VMs and containers promises operational consistency and cost savings, it also presents significant challenges, including a skills gap for VM operators transitioning to Kubernetes. Projects like KubeVirt and Red Hat's OpenShift...
Source: The New Stack
Dean Lewis

Technical Debt vs. Architecture Debt: Don’t Confuse Them

2025-10-22 13:00
Understanding the difference between technical debt and architecture debt is crucial for organizations. Technical debt refers to visible issues like bugs or code flaws that can be addressed relatively easily. In contrast, architecture debt is systemic and often unnoticed until major projects stall, affecting overall performance. This distinction can save companies from significant costs. The article emphasizes recognizing these types of debt to ensure smoother project execution and long-term...
Source: The New Stack
Nadzeya Stalbouskaya

An Ethics Crash Course for Agentic AI: Autonomy Versus Trust

2025-10-22 12:05
🚗 Today’s leaders are navigating the complexities of AI with self-driving capabilities. The article discusses the balance between AI autonomy and trust. Organizations need to define their strategy: should AI operate independently or under human supervision? This is crucial for sustainable deployment and regulatory compliance. Understanding the spectrum of AI autonomy, from human-in-the-loop to fully autonomous systems, helps in making informed decisions. Explore how the six pillars of...
Source: The New Stack
Vrushali Sawant

As Vibe Coding Fades, Woz Offers Production-Ready Alternative

2025-10-21 22:00
As the app-building bubble deflates, experts Ben Collins and Brad Eckert highlight the limitations of current rapid prototyping tools. 🚫📉 Collins, CEO of the new AI platform Woz, notes that many tools generate unusable code, leading to wasted time and resources. Woz aims to bridge this gap by combining automation with human engineers to create production-ready apps. They emphasize the need for effective solutions that businesses, especially without large engineering teams, can rely on....
Source: The New Stack
Darryl K. Taft

React Foundation Leader on What’s Next for the Framework

2025-10-21 18:05
🚀 Exciting updates from the React Foundation! Seth Webster, the foundation's first executive director, shared plans for better bootcamps and support for underserved communities. A merchandising store is also in the works, responding to high demand for React-themed items. 🧢👕 The React Foundation, launched on Oct. 7 under The Linux Foundation, aims to expand React and React Native's impact beyond Meta. Despite shifting focus to AI, Meta remains committed to supporting React for the long term....
Source: The New Stack
Loraine Lawson

How Thrivability ROI Tackles the Cognitive Load Crisis

2025-10-21 17:00
In the evolving software development landscape, complexity poses a significant challenge. As tools become more sophisticated, the cognitive load on development teams increases, impacting innovation and productivity. 💻⚙️ This cognitive overload is costing businesses an estimated $322 billion annually, highlighting its importance as a financial concern. To address this, the concept of "Thrivability ROI" is introduced, focusing on reducing cognitive load to enhance developer effectiveness and...
Source: The New Stack
Betty Junod

Should You Go All-In on Vite? A Risk vs. Reward Analysis

2025-10-21 16:00
🚀 Vite has emerged as more than just a JavaScript bundler; it's become essential for many modern frameworks. With fast startup times and instant hot module replacement, Vite enhances developer experience significantly. However, going all-in on Vite means depending on a single approach in a competitive landscape. Explore the benefits and potential risks of embracing Vite fully in this detailed analysis. #Vite #JavaScript #WebDevelopment #DeveloperExperience #FrontendFrameworks
Source: The New Stack
Alexander T. Williams

Beyond Namespaces: Why Kubernetes Needs Real Workload Isolation

2025-10-21 15:00
Kubernetes namespaces are essential for managing resources, but they provide only logical separation, not true workload isolation. 🛠️ This means that while namespaces help teams share clusters, they do not prevent compromised containers from affecting others on the same node. Real security is needed to protect against vulnerabilities and potential breaches. 🔒 Modern attack patterns highlight the risks, showing that relying solely on namespaces can lead to significant security issues. ⚠️ For...
Source: The New Stack
Lewis Denham-Parry

Building a Self-Aware Enterprise With GenAI

2025-10-21 14:00
🚀 The concept of a self-aware enterprise is evolving with advancements in generative AI (GenAI). Enterprises are moving towards accessing all organizational data without pre-set limitations, allowing for deeper analysis and insights. Key innovations include the commercialization of large language models and scalable vector data stores, which enhance data accessibility and semantic search capabilities. Transitioning to this model requires careful planning and teamwork, emphasizing simplicity...
Source: The New Stack
Ed Huang

How To Get Bare-Metal GPU Performance in Confidential VMs

2025-10-21 11:00
At OpenInfra Summit Europe 2025, NVIDIA highlighted the importance of combining Kata Containers with Confidential Computing for running sensitive AI workloads on GPUs. 🖥️🔒 Kata Containers offer lightweight VMs that provide strong isolation while allowing container performance. Confidential Computing adds encryption for in-memory data, ensuring even cloud providers cannot access sensitive information. 🔐 NVIDIA's Zvonko Kaiser emphasized the need for attestation to verify the VM's state before...
Source: The New Stack
Steven J. Vaughan-Nichols

A Cascade of Failures: A Breakdown of the Massive AWS Outage

2025-10-21 00:00
🚨 AWS experienced a significant outage in its US-EAST-1 region, affecting multiple services like AWS Lambda and Amazon API Gateway. The issues began around 3 a.m. EDT due to DNS misconfiguration, leading to increased error rates. By 6 a.m., recovery efforts were underway, but challenges with EC2 instances persisted. As of the latest update, services have largely recovered, though a backlog remains for some users. #AWS #CloudServices #Outage #TechNews #DNS
Source: The New Stack
Joab Jackson

Why Google Cloud’s OTLP Support Matters for Your Telemetry Pipeline

2025-10-20 22:00
Google Cloud has enhanced its Observability features by enabling Cloud Trace to support the OpenTelemetry Protocol (OTLP) for trace data ingestion. This allows users to send trace data directly via the Telemetry API, making it easier for those handling large volumes of data. 📊 The integration simplifies the telemetry pipeline and reduces the need for vendor-specific exporters, promoting standardization across various observability backends. This move aligns Google Cloud's approach with...
Source: The New Stack
B. Cameron Gain

How Block Got 12,000 Employees Using AI Agents in Two Months

2025-10-20 21:00
🚀 Block has successfully deployed AI agents to its 12,000 employees in just eight weeks, a feat shared at the All Things Open 2025 conference. It began with a single engineer frustrated by limited AI capabilities. After experimenting with OpenAI’s function calling, Block adopted the Model Context Protocol to enhance automation. The team pivoted to focus on user-friendly design, ensuring even non-technical staff could easily access the tools. #AI #Innovation #TechNews #EmployeeEngagement #Block
Source: The New Stack
Darryl K. Taft

Why Modern IPv6 Failed This Massive Kubernetes Networking Test

2025-10-20 19:00
Deutsche Telekom is tackling the challenge of simulating dynamic satellite networks, including those like SpaceX's Starlink. 🚀 At the OpenInfra Summit Europe 2025, experts shared insights on their efforts to create a scalable, container-based testbed. Their record-breaking Kubernetes cluster runs 2,000 pods with 10,000 network interfaces. 🌐 This achievement highlights the difficulties of emulating networks with moving nodes, where traditional routing protocols struggle. The project aims to...
Source: The New Stack
Steven J. Vaughan-Nichols

Anthropic’s Claude Code Comes to Web and Mobile

2025-10-20 18:00
🚀 Anthropic's Claude Code is now available on the web and mobile! Developers can start workflows from the Claude.ai app, allowing for multi-tasking with coding tasks. 🔒 Each session runs in a secure environment, ensuring safe access to GitHub repositories. Users can assign tasks while monitoring progress. ✨ The new real-time steering feature lets developers guide Claude without interruption. This means more efficiency and less hands-on management. #ClaudeCode #Anthropic #WebDevelopment #AI...
Source: The New Stack
Frederic Lardinois

Technical Leadership for Business Impact: Focus on Customers

2025-10-20 17:00
Every growing company faces the challenge of scaling technical leadership effectively. 📈 CTOs must adapt their approach to meet shifting business demands and market conditions. It's essential for leaders to evolve while maintaining a focus on technical excellence, business strategy, and customer success. There are three operational models for CTOs: Builder, Strategist, and Guardian. Each plays a unique role in aligning technical efforts with customer needs. Understanding these models can help...
Source: The New Stack
Sabrina Farmer

MCP vs. API Gateways: They’re Not Interchangeable

2025-10-20 15:00
Organizations are increasingly adopting the Model Context Protocol (MCP) to link their services and data with AI models. However, they face challenges in securing access to MCP servers while ensuring effective routing and observability. 🔐 A common question arises: Can existing API gateways be used for MCP? The answer is complex. API gateways were designed for stateless interactions, while MCP operates with stateful sessions. This fundamental difference means that current API gateways may need...
Source: The New Stack
Christian Posta

Make Data Ready for AI With Hygiene, Governance, and Experimentation

2025-10-19 17:00
Is your data ready for AI? 🤖 As organizations plan for AI adoption, many face challenges due to data quality rather than model limitations. The article highlights that effective AI relies on clean, governed, and context-rich data. Data access issues lead to stalled projects, showing the importance of hygiene, governance, and experimentation. 🔍 Key solutions include data federation for better access and iceberg data lakehouses for scalability. These approaches enhance experimentation speed and...
Source: The New Stack
Brian Luisi