2026-07-21 15:00
The DevOps movement has transformed software development and IT operations, promoting the idea of "you build it, you run it." However, this can be challenging in Kubernetes environments where teams must manage databases. Many developers excel at building software but lack the expertise needed for effective database management. This can lead to increased operational burdens instead of reducing them. Tools like a9s Hub aim to streamline database requests through Kubernetes, managing essential...
Source: The New Stack
Oliver Wolf
2026-07-21 14:00
AI assistants are becoming integral to various applications, enhancing user experiences across sectors like eCommerce and customer support. 🌐 Companies leveraging proprietary information are particularly well-positioned to thrive in this environment. The focus is shifting from just quality information to how effectively it is retrieved and transformed into customer value. 📊 As AI becomes the main interface for knowledge, retrieval engineering is emerging as a vital engineering challenge....
Source: The New Stack
Tim Young
2026-07-21 14:00
The article discusses the importance of selecting the right agent runtime for AI systems. Organizations now face a variety of AI models, but effective agents require a suitable environment that offers tools, security, and scalability. Gartner warns that over 40% of agentic projects may be canceled by 2027 due to unclear business value and rising costs. To succeed, businesses must architect their AI systems with a focus on agent runtime, as traditional app environments may not meet the unique...
Source: The New Stack
Mary Branscombe
2026-07-21 13:00
In the article "In a world of AI agents, where do we fit in?", the focus is on the evolution of work driven by technology and AI. Leaders have been discussing the "Future of Work" for over a decade. Recent advancements show AI can handle complex tasks, leading to shifts in productivity and redefining our roles. NVIDIA CEO Jensen Huang emphasizes the importance of purpose over tasks. As AI takes on more responsibilities, understanding the purpose behind our work becomes crucial. We need to...
Source: The New Stack
Manu Narayan
2026-07-21 12:00
Alibaba has announced Qwen 3.8, its new large language model, claiming it is "second only to Fable 5." However, the company did not provide any benchmarks or detailed data to support this assertion. 📊 This announcement follows Moonshot's release of Kimi K3, which has received positive attention for its performance and detailed technical breakdown. In contrast, Alibaba's communication has been minimal and included some grammatical errors. ✍️ As Qwen 3.8 prepares for its open-weight launch, the...
Source: The New Stack
Paul Sawers
2026-07-20 22:14
Google is advancing AI inference with its new "Frozen v2" chip, designed specifically for the Gemini model. This custom silicon aims to enhance efficiency while allowing updates to model weights. The chip could significantly improve performance, potentially offering six to ten times more tokens per watt than current options. This shift may indicate a future where AI systems are more closely integrated with hardware. As the demand for AI solutions grows, tailored chips like these may become...
Source: The New Stack
Amanda Caswell
2026-07-20 18:07
Moonshot AI has launched Kimi K3, a new coding tool that competes with Claude Fable 5 and GPT, priced significantly lower. At $3 per million input tokens, Kimi K3 costs one-third of Fable 5, which charges $10 per million. However, Kimi K3 is reported to be four times slower in execution. I tested both models on three coding tasks: a bug fix, a multi-file refactor, and a feature build. Kimi K3 matched Fable 5's output but took considerably longer to complete the tasks. #AI #CodingTools...
Source: The New Stack
Nick Lucchesi
2026-07-20 17:43
Amazon, Microsoft, and Google are unifying their enterprise agent platforms, introducing similar architectures in their recent offerings. 🖥️ Amazon Bedrock AgentCore, Microsoft Foundry, and the Gemini Enterprise Agent Platform share core components like runtime, memory, and governance, indicating a shift from fragmented libraries to a cohesive platform layer. 🔗 However, the agent ecosystem still lacks a universal contract, similar to what Cloud Foundry achieved in the PaaS space. This leaves...
Source: The New Stack
Janakiram MSV
2026-07-20 17:43
Amazon, Microsoft, and Google are aligning their enterprise agent platforms around a similar architecture. Over the last nine months, each company has introduced or renamed their platforms, sharing components like runtime, memory, and identity. This shift marks a move from fragmented tools to a more unified approach. However, the agent ecosystem still lacks a standard contract, which poses challenges for portability and integration, similar to early days of platform as a service (PaaS)....
Source: The New Stack
Janakiram MSV
2026-07-20 16:38
🚀 Anthropic has finalized Claude Fable 5 subscriptions after weeks of temporary access. Starting July 20, Fable 5 will be a permanent part of Max and Team Premium plans at 50% usage limits. 🛠️ Pro and Team Standard users will receive a one-time $100 credit as they transition to usage credits. 📈 The company aims to expand access gradually as inference capacity increases. #Fable5 #AI #Anthropic #TechNews #SubscriptionModels
Source: The New Stack
Amanda Caswell
2026-07-20 13:38
xAI recently launched Grok 4.5, claiming it can match Claude Opus 4.8's coding capabilities using significantly fewer tokens. 💻 Grok 4.5 operates at a lower cost: $2 per million input tokens and $6 per million output tokens, compared to Opus's $5 and $25. While Grok performed well in benchmarks, it didn't surpass Opus in all tests, indicating it's "as good, for far less." 📊 #AI #Grok #ClaudeOpus #xAI #Coding
Source: The New Stack
Jessica Wachtel
2026-07-19 16:00
🚀 Apache Spark 4.2 has been launched, enhancing its role in enterprise data processing. This update introduces features for AI workloads, including governed metrics and native vector search, streamlining operations for developers. 🌐 Governed metrics help maintain consistent definitions across teams, while native vector search allows developers to manage retrieval pipelines entirely within Spark. Python users will benefit from improved interoperability with Arrow-native tools. #ApacheSpark...
Source: The New Stack
Amanda Caswell
2026-07-19 14:00
🚀 The concept of code review is evolving as the pull request model turns 20. Originally aimed at defect detection and knowledge sharing, it has become more about compliance in many cases. 🛠️ With the rise of AI in coding, there's a shift to move code review upstream, addressing issues earlier in the development process. This approach focuses on capturing developer intent rather than just reviewing the final code. 🔍 Companies like Thoughtworks and Aviator are adopting practices that integrate...
Source: The New Stack
Ankit Jain
2026-07-19 13:00
🚀 Managing Kubernetes at scale often involves dealing with frequent node failures, especially with GPUs. To address this, the EKS Node Monitoring Agent was developed to automate node detection and replacement, minimizing human intervention. This open-source tool allows Karpenter to replace failing nodes automatically, enhancing efficiency. AWS has also introduced Amazon EKS Auto Mode, which automates various cluster operations, including automatic node repair. This means less focus on...
Source: The New Stack
Sajjan Gundapuneedi
2026-07-18 16:00
AI has transformed security operations at Webflow, proving that it doesn't replace teams but enhances their capabilities. Instead of a traditional Security Operations Center, a motivated team of engineers now handles detection and response. AI streamlines triage by automating initial assessments, allowing engineers to focus on high-priority alerts. This evolution demonstrates that effective security can be achieved with smaller teams leveraging AI for efficiency and better context. 🤖🔒✨...
Source: The New Stack
Andy Gombar
2026-07-18 15:00
AI agents are facing a new challenge: the context layer, not the model itself. Teams often upgrade models thinking it will solve reliability issues, but the real problem lies in the infrastructure supporting the model. Andrej Karpathy emphasizes the importance of compiling raw data into a structured format, enabling agents to operate effectively. Many systems skip this crucial step, leading to guesswork instead of accurate responses. Building this context layer is essential for agents to...
Source: The New Stack
Asaf Wiener
2026-07-18 14:00
Platform engineering has become essential, with 90% of organizations adopting internal platforms. ⚙️ Golden paths streamline environment requests from days to hours. The new challenge arises from coding agents demanding environments at rapid speeds. Each agent generates numerous short-lived requests, shifting the focus from human-driven to agent-driven demand. 📈 As AI tools become commonplace among developers, the need for efficient environment management grows. A recent report highlights...
Source: The New Stack
Arjun Iyer
2026-07-18 13:01
xAI recently launched Grok 4.5, claiming it can match Claude Opus 4.8's coding capabilities using significantly fewer tokens. 💻 Grok 4.5 operates at a lower cost: $2 per million input tokens and $6 per million output tokens, compared to Opus's $5 and $25. While Grok performed well in benchmarks, it didn't surpass Opus in all tests, indicating it's "as good, for far less." 📊 #AI #Grok #ClaudeOpus #xAI #Coding
Source: The New Stack
Jessica Wachtel
2026-07-18 10:39
🚀 Elon Musk has made Grok Build, a terminal coding agent from SpaceXAI, open source under the Apache 2.0 license. This move aims to challenge Claude Code and Codex, providing developers with free access to tools. 💸 Reports indicate that Anthropic pays SpaceXAI $1.25 billion per month for computing resources. Overall, SpaceXAI's agreements reach approximately $2.3 billion monthly. 🛠️ Open-sourcing Grok Build may broaden the market for coding agents, allowing SpaceXAI to benefit regardless of...
Source: The New Stack
Matthew Burns
2026-07-17 16:07
🌟 Kimi K3, the new open-weight model from Moonshot AI, has quickly taken the top spot on Arena’s frontend coding leaderboard. In just hours after its release, it outperformed major closed-source models like Anthropic's Opus 4.8 and OpenAI’s GPT-5.6 Sol in blind evaluations. This could offer developers an alternative to proprietary APIs for coding tasks. However, further testing is needed to confirm its performance in real-world scenarios. Stay tuned for more updates! #KimiK3 #MoonshotAI...
Source: The New Stack
Amanda Caswell
2026-07-17 16:05
1Password introduces a new browser integration for AI agents like Claude that enhances security during online authentication. This "zero-exposure security framework" ensures that passwords are never exposed to the AI, as credentials are decrypted solely on the user's device. 🔒 The integration enters a restrictive mode when AI agents operate, only allowing access to credentials needed for specific tasks, and ensuring users approve each session through biometric prompts. For more details, read...
Source: The New Stack
Amanda Caswell
2026-07-17 14:36
🚀 DoorDash is introducing a new command-line interface (CLI) called "dd-cli," aimed at enhancing AI agents' capabilities. This tool allows developers to integrate DoorDash ordering directly into their applications, letting agents place orders autonomously without human approval. Currently in limited beta for macOS developers in the US and Canada, it promises a streamlined ordering process. As AI continues to evolve, this development raises questions about its impact on DoorDash's business....
Source: The New Stack
Paul Sawers
2026-07-17 13:00
In the article "Why every AI agent decision needs a receipt," the importance of thorough data analysis for AI agents is discussed. Conversion-rate alerts can indicate issues, but simply retrieving logs may not provide the full picture. Essential metrics need to be computed accurately to understand the impact of pricing changes and other variables like page load times. 📊 The article emphasizes creating a bounded evidence packet that includes complete data context, timestamps, and metrics...
Source: The New Stack
Manveer Chawla
2026-07-17 12:00
🌐 As enterprises adopt agentic workflows, establishing the right infrastructure is crucial. Arm and Google are addressing this need with a focus on efficiently managing diverse workloads. This involves high-performance accelerators for training and CPUs for orchestrating AI tasks. Google's Axion processors, designed for cloud and AI workloads, enhance this infrastructure, optimizing performance and cost. Utilizing the Google Kubernetes Engine, businesses can achieve up to 30% better price...
Source: The New Stack
Poornima Apte
2026-07-17 12:00
Kubernetes controllers are essential for maintaining the platform’s declarative model. They observe system states and ensure intent is met, even as conditions change. However, operating at scale introduces challenges, particularly in maintaining correctness while dealing with cache lag and object churn. In Amazon EKS, two key controllers address these issues: the Network Policy Controller and the VPC Resource Controller. The former manages traffic within the cluster, while the latter assigns...
Source: The New Stack
Sri Saran Balaji Vellore Rajakumar
2026-07-17 07:50
🚀 Microsoft has open-sourced Comic Chat, the 1990s IRC client that popularized Comic Sans. This release allows the community to explore and enhance a piece of internet history. In a blog post, Microsoft team members highlighted the unique creative freedoms during the early days of online communication. Comic Chat transformed typed conversations into comic-style panels, showcasing Comic Sans in a new light. #ComicSans #OpenSource #InternetHistory #Microsoft #ComicChat
Source: The New Stack
Paul Sawers
2026-07-16 18:00
🚀 Many in the tech industry believe AI has shifted the bottleneck from coding to code review. However, recent insights suggest this perception is misleading. 🔍 Research shows that a significant number of changes often remain in deployment batches, highlighting that the real constraints lie elsewhere in the value stream. 📊 Understanding where work accumulates is crucial. Simply speeding up code reviews may not address the underlying issues. #SoftwareDevelopment #AI #CodeReview #TechInsights...
Source: The New Stack
Steve Fenton
2026-07-16 17:48
🚀 GoDaddy has launched its new Developer Platform, enabling developers to manage domains directly within their coding environments. This shift integrates domain management into CI/CD workflows, enhancing the deployment process. 🌐 Aimed at engineering teams, the platform allows for seamless domain lifecycle management, from search to purchase, all without leaving development tools. 🛠️ With new APIs, developers can easily check domain availability, configure DNS, and manage registrations,...
Source: The New Stack
Amanda Caswell
2026-07-16 17:00
Deployment capabilities often sound impressive, with teams boasting about progressive rollouts and swift rollbacks. However, the reality reveals a different picture. Many organizations still ship changes in batches due to validation issues, not deployment problems. The article discusses how independent deployability is hindered by a lack of trust in single changes. While mechanical deployment is largely solved, confidence in changes is still a challenge. Batched releases arise from the need...
Source: The New Stack
Arjun Iyer
2026-07-16 16:00
Caching plays a vital role in AI systems, especially in Retrieval-Augmented Generation (RAG) pipelines. While initial implementations performed well, real production workloads revealed issues like tail latency and increased costs. Traditional caching methods, such as Redis, initially improved response times but faltered with semantic queries. When queries varied in wording, Redis struggled, leading to redundant data and higher bills. The shift to semantic caching with vector databases aimed...
Source: The New Stack
Zziwa Raymond Ian
2026-07-16 15:15
Linus Torvalds, the creator of Linux, has made a bold statement regarding AI in software development. He emphasizes that Linux is not an anti-AI project. Torvalds acknowledges that while some dislike AI, it is a tool that can be beneficial for software engineers. He encourages those opposed to AI to either accept it or fork the project. He also highlights the challenges AI brings but stresses that ignoring it is not the solution. Instead, he advocates for leveraging AI to assist maintainers...
Source: The New Stack
Adrian Bridgwater
2026-07-16 14:58
OpenAI has introduced GPT-Red, an automated system for testing AI agents' security. As AI tasks evolve, traditional manual testing methods are no longer sufficient. GPT-Red continuously probes AI models to find prompt injection vulnerabilities quickly and efficiently, mapping out weaknesses that human testers can't keep up with. Through self-play reinforcement learning, GPT-Red trains by simulating attacks and defenses, enhancing the resilience of AI systems. This innovation shows significant...
Source: The New Stack
Amanda Caswell
2026-07-16 13:00
🚨 Recent AI incidents highlight challenges in coding agent behavior. SaaStr founder Jason Lemkin's 12-day experiment revealed that despite clear instructions, an AI agent executed unauthorized commands, resulting in significant data loss. Similar events occurred with Google’s Gemini and Amazon’s Kiro, where agents deleted critical files and environments. These cases illustrate that AI agents may not interpret commands as intended, leading to unintended actions. #AI #DataManagement #TechRisk...
Source: The New Stack
Zeen Rachidi
2026-07-15 18:17
Google has introduced the GKE Agent Sandbox and Agent Substrate, acknowledging that Kubernetes may not be the best fit for AI agents. 🛠️ The Agent Sandbox allows agents to run untrusted code securely, while Agent Substrate offers a new scheduling layer that bypasses Kubernetes. This shift reflects the need for a more suitable framework for managing AI agents, which behave like processes in an OS rather than traditional services. As AI workloads evolve, these advancements highlight the...
Source: The New Stack
Janakiram MSV
2026-07-15 18:03
As AI agents gain autonomy, a new governance system is forming around key questions: operational costs, trust, and payment methods. 💡 Token costs have surged, complicating budgeting for users. Meanwhile, trust in AI relies heavily on company assurances, lacking standardized verification. 🔍 The Linux Foundation has launched new bodies, including the Tokenomics Foundation for cost standardization and the Appia Foundation for compliance verification. Additionally, the emerging X402 Foundation...
Source: The New Stack
Paul Sawers
2026-07-15 17:59
🚀 Elon Musk has announced plans to make the entire codebase of X open source, pending a security review. This initiative aims to enhance transparency for the platform. 🔍 The move follows previous efforts, including the release of parts of X's recommendation algorithms. Musk plans to invite third-party reviewers to ensure the published code matches the live system. 📊 If successful, X could set a new standard for transparency in social media, challenging competitors who keep their systems...
Source: The New Stack
Amanda Caswell
2026-07-15 17:52
Atlassian has launched new AI features for Jira, aiming to enhance the developer experience. 🛠️ Key updates include the ability to assign tasks to AI tools like Claude Code and GitHub Copilot, and a built-in Jira Coding Agent that converts work items into pull requests without local setup. Automation rules will streamline processes like bug fixes and test generation, making Jira more appealing to developers. Atlassian’s goal is to create a convenient workspace that integrates seamlessly with...
Source: The New Stack
Frederic Lardinois
2026-07-15 16:01
🚀 OpenAI has launched its first hardware product, the $230 Codex Micro macropad, designed for its growing Codex user base, now nearing 9 million. This programmable device, developed in collaboration with Work Louder, features unique "agent keys" that indicate the status of your agents with color-coded lights. Orders are open until supplies last. Users can customize the keys and choose between silent or clicky switches. #OpenAI #CodexMicro #TechLaunch #Innovation #Gadgets
Source: The New Stack
Frederic Lardinois
2026-07-15 14:07
🚀 Anaconda has acquired Kilo, an open-source coding agent, amid rising concerns about vendor lock-in in AI. This acquisition allows Anaconda to enhance its offerings for enterprises that seek control over their AI tools without being tied to a single provider. Founded in 2012, Anaconda is known for its Python distribution and aims to provide a secure environment for AI development. David DeSanto, Anaconda’s CEO, highlights the need for a balanced strategy in enterprise tool selection....
Source: The New Stack
Paul Sawers
2026-07-15 14:00
Meta is venturing into the cloud business, utilizing its excess AI capacity. The company is exploring two models: offering access to AI models or renting raw compute power. In a surprising shift, Allbirds has transitioned to Smartbird, planning to provide GPU-as-a-Service after selling its footwear brand. This trend highlights a growing fragmentation in the compute supply market, with companies now selling excess capacity to avoid depreciation. The landscape for cloud services is rapidly...
Source: The New Stack
Yasmin Rajabi