2026-04-27 18:53
In a recent episode of The New Stack Makers, Nimisha Asthagiri, data and AI advisor at Thoughtworks, discussed the challenges companies face in advancing AI from proof of concept to production. Many organizations are struggling with this transition, as Gartner predicts over 40% of agentic AI projects may be canceled by 2027. Asthagiri emphasizes the need for companies to rethink their approach to AI, focusing on what new capabilities can be built instead of just speeding up processes. The...
Source: The New Stack
Frederic Lardinois
2026-04-27 16:08
Executives are increasingly "vibe coding" their own tools using AI coding assistants, moving beyond traditional IT processes. This trend includes simple automations to complex systems, driven by impatience and curiosity. Tools like Claude and Cursor are popular among high-level professionals. Moshe Bar, CEO of Codenotary, successfully created a bulletin board system with minimal coding, demonstrating the potential of AI in software development. #AICoding #BusinessInnovation #TechTrends...
Source: The New Stack
Darryl K. Taft
2026-04-27 15:00
🚀 The 2026 State of Open Source Report highlights a significant shift in digital autonomy discussions among CIOs and CTOs. As organizations integrate open source into their systems, concerns about vendor lock-in have risen, with 68% citing it as a primary adoption driver. This trend emphasizes the importance of control and flexibility in tech environments. However, increased autonomy comes with challenges. Many enterprises report that over 50% of engineering time is now spent on maintenance...
Source: The New Stack
Matthew Weier O’Phinney
2026-04-27 14:00
The demand for AI applications is growing rapidly, with many developers focusing on the need for structured data. Jensen Huang of Nvidia emphasized that structured data is the "ground truth for AI." This is crucial for preventing AI agents from making inaccurate assumptions. Postgres has emerged as the preferred database for AI workloads. According to the Stack Overflow 2025 Developer Survey, 66% of developers prefer Postgres for its ease of use and strong community support. #Postgres #AI...
Source: The New Stack
Meredith Shubel
2026-04-27 13:00
Security excellence is critical for software companies, linking directly to software quality. However, aligning security ambitions with operational reality presents challenges. Many organizations find that new security tools do not integrate well with existing systems, leading to inefficiencies. This can turn security into a bottleneck instead of an enabler. To succeed, companies should focus on operational maturity, which includes modern architecture, automated deployment processes, and a...
Source: The New Stack
Jamie Dicken
2026-04-27 12:00
🚀 Exciting news from Brian Douglas, a GitHub veteran! He has launched Paper Compute to address the infrastructure gap in AI agent development. Douglas emphasizes the need for robust systems beneath AI agents, which are often improvised. Paper Compute aims to provide a cohesive solution built on open source principles. Their first product, Tapes, offers observability for agent activity without altering application code. This is designed to enhance confidence in deploying agents across various...
Source: The New Stack
Paul Sawers
2026-04-26 17:16
🔍 In a significant shift, OpenAI and DeepSeek have made contrasting moves in the AI market. OpenAI launched GPT-5.5 with increased pricing, while DeepSeek introduced more affordable options with V4-Pro and V4-Flash. This creates a widening gap in AI pricing, impacting developers' choices. The middle tier is thinning, pushing coders to reassess their options. 📉🤖 #AI #OpenAI #DeepSeek #MachineLearning #TechTrends
Source: The New Stack
Janakiram MSV
2026-04-26 16:00
Building KubeStellar Console with AI agents revealed unexpected challenges. Initially, the coding agents produced quick results, making progress feel effortless. However, issues arose with builds breaking and scope expanding unexpectedly, leading to frustration. The key to success was not in the AI model itself, but in how the surrounding codebase learned to measure and respond. After four months, PR acceptance reached 81%, thanks to the implementation of five maturity loops. #KubeStellar #AI...
Source: The New Stack
Andy Anderson
2026-04-26 15:00
A recent Slack message sparked an important discussion within an engineering team: "What are we actually running across both cloud environments?" 🤔 Despite being a straightforward question, it revealed gaps in their understanding of their multi-cloud infrastructure. It took an hour and several replies to gather information, leading to a realization that a cohesive map of their stack was missing. 📊 As organizations often accumulate cloud services over time without a clear strategy, many find...
Source: The New Stack
Joe Karlsson
2026-04-26 14:00
Unlocking the data within videos is becoming increasingly important as the volume of video content grows. 🎥 Businesses can leverage Video Management Systems to search through videos for valuable information, from e-commerce product detection to academic lectures. 📚 However, challenges remain. Video data is complex, with visual and audio elements processed separately, making it tricky to extract meaningful content efficiently. #VideoData #DataExtraction #VideoManagement #Ecommerce...
Source: The New Stack
Kai Borgen
2026-04-26 13:00
Cursor 3 has launched with a new feature: the Agents Window. 🖥️ This interface allows users to describe tasks for AI agents to execute. The design resembles Claude Code's interface, indicating a competitive response to Anthropic’s offerings. The article discusses tests comparing debugging capabilities of Cursor 3 and Claude Code, specifically focusing on a security bug in HTTPie. 🔍 Both tools were prompted to fix a documented issue involving terminal escape sequences. The results of these...
Source: The New Stack
Jessica Wachtel
2026-04-25 16:00
Jaeger is evolving to meet the demands of modern software architectures. With the rise of microservices and generative AI, observability tools must adapt. The introduction of Jaeger v2 integrates OpenTelemetry, enhancing data collection for AI workloads. This update supports collaboration between engineers and AI agents, streamlining the tracing of complex execution paths. Additionally, Jaeger is adopting new open standards to further improve interaction and debugging processes. 🌐📊🤖 #Jaeger...
Source: The New Stack
Jonah Kowall
2026-04-25 10:14
OpenAI's recent announcement highlights the launch of Workspace Agents, a significant development in enterprise AI. This new feature allows organizations to create and share AI agents across teams, enhancing collaboration and efficiency. Admins can govern access and tools for each user group, making it a powerful infrastructure for businesses. While GPT-5.5 received attention, Workspace Agents may be more impactful for users seeking to streamline workflows. 🌐🤖📊 #AI #OpenAI #WorkspaceAgents...
Source: The New Stack
Matthew Burns
2026-04-24 16:04
Mistral AI has launched Leanstral, an open-source code agent aimed at reducing the need for human-in-the-loop code checks in software engineering. Leanstral utilizes formal verification and the Lean 4 programming language to ensure code performs as specified, minimizing bugs. 🌐 With a focus on high-stakes sectors like aerospace and cryptography, Leanstral has a unique architecture that enhances efficiency. Developers can access it via a free API. However, the effectiveness of Leanstral in...
Source: The New Stack
Adrian Bridgwater
2026-04-24 13:42
🌐 Vectors transformed AI search by converting data into numerical formats, enabling semantic search and recommendations. However, their one-dimensional nature limits effectiveness. 🔍 Enter tensors: multi-dimensional structures that enhance search capabilities with improved ranking and context. This means smarter AI that can handle complex data and longer documents. 📅 Join a discussion with industry experts on May 5 to explore how tensors can future-proof AI applications in various sectors....
Source: The New Stack
Alex Wilhelm
2026-04-24 12:49
OpenAI has launched its latest models, GPT-5.5 and GPT-5.5 Pro, boasting improvements in coding and complex task execution. Despite its consumer success, developers show a preference for Anthropic's Claude models. 🤖 Benchmark results highlight advancements, yet real-world performance remains a crucial query. Early tester Simon Willison noted limitations like the absence of API access, affecting his evaluations. 🛠️ While some tests showed slower output compared to GPT-5.4, adjustments yielded...
Source: The New Stack
Paul Sawers
2026-04-24 12:00
Cursor and Chainguard have partnered to enhance security in the open source dependency chain for AI-generated code. This collaboration aims to protect against supply chain attacks that have become more prevalent as AI development scales up. 🔒🤖 With this partnership, Cursor can now access Chainguard’s verified container images and language libraries, allowing for safer dependency selection directly in the coding workflow. This change addresses recent threats targeting public registries. 📦 As...
Source: The New Stack
Darryl K. Taft
2026-04-24 12:00
Boris Cherny recently highlighted the importance of validating cloud-native code with Claude Code following the release of Opus 4.7. He emphasizes that giving Claude a way to verify its work can significantly enhance productivity, especially in complex cloud environments. This practice has become a standard for coding agents. Major players like OpenAI's Codex and GitHub's Copilot have implemented similar validation loops, ensuring that agents check their work before submission. This trend...
Source: The New Stack
Arjun Iyer
2026-04-23 20:54
OpenAI has launched the Privacy Filter, a new tool for detecting and redacting personally identifiable information (PII) on local devices. This bidirectional token-classification model scans long-form text in one pass, offering greater context awareness and privacy for users. It can identify and mask various types of PII, including names, addresses, and emails, while running efficiently on laptops. Privacy Filter aims to improve upon traditional PII detection methods by understanding context,...
Source: The New Stack
Meredith Shubel
2026-04-23 18:39
OpenAI has introduced shared "workspace agents" to enhance team operations with ChatGPT. These agents can manage multi-step tasks like drafting reports and responding to messages, operating autonomously without constant input. They integrate with existing tools and can be used across teams, streamlining workflows. Teams can build these agents by outlining workflows in ChatGPT, enabling automation and efficiency. Custom GPTs will still be available, with plans for future conversions to agents....
Source: The New Stack
Paul Sawers
2026-04-23 18:04
🚀 OpenAI has launched GPT-5.5 and GPT-5.5 Pro, marking a significant advancement in AI capabilities. 💡 The new models are designed for all paying users in ChatGPT and Codex, with Pro features available for Business and Enterprise users. 📈 Early benchmarks show GPT-5.5 outperforms competitors, especially in coding tasks, while maintaining speed and efficiency. 💬 OpenAI's Greg Brockman describes it as “a new class of intelligence” that enhances intuitive computing. #OpenAI #GPT55...
Source: The New Stack
Frederic Lardinois
2026-04-23 17:48
🚀 Kyverno recently graduated from the CNCF, marking a significant achievement for the open-source policy engine for Kubernetes. In an interview, Jim Bugwadia, co-founder and CEO, highlighted that reaching this level took five years of growth and stability. Kyverno offers essential features like centralized fleet management and observability for large-scale Kubernetes deployments. The user base is expanding, driven by the increasing need for policy enforcement amidst rising AI workloads. With...
Source: The New Stack
Jennifer Riggins
2026-04-23 17:04
Only 5% of firms report seeing ROI from AI investments. Despite high expectations for productivity gains, many organizations are struggling to achieve results. Join us on May 7 at 2:30 PM ET for a discussion with experts from Octopus Deploy. We’ll explore findings from the AI Pulse Report and share strategies to improve AI metrics. Don’t miss out—register now for insights and a chance to win Octoshop swag! 🎉 #AI #Productivity #TechTalk #Webinar #AIAdoption
Source: The New Stack
Jennifer Riggins
2026-04-23 12:52
Anthropic's Claude Opus 4.7 was launched with promises of enhanced reasoning and instruction-following capabilities. However, users are reporting challenges with the model's performance. Some prompts that worked with previous versions now yield unexpected results. Users are finding that Claude 4.7 often second-guesses itself, leading to inefficiencies in complex tasks. Concerns from the community highlight that despite being a premium service, many feel the model has not met expectations. #AI...
Source: The New Stack
Adrian Bridgwater
2026-04-22 18:43
🚀 SpaceX has announced a significant partnership with AI startup Cursor, valued between $10 billion and $60 billion. This collaboration aims to develop advanced coding and knowledge work AI by combining SpaceX’s resources with Cursor's successful AI coding tools. Cursor has seen impressive growth and access to valuable usage data from developers, while SpaceX’s xAI seeks to enhance its AI capabilities. This partnership could lead to stronger AI models and a more competitive edge in the...
Source: The New Stack
Alex Wilhelm
2026-04-22 18:07
At the MCP Summit in NYC, AWS's Luca Chang discussed the Model Context Protocol (MCP), which is becoming essential for connecting AI models with tools and data. He highlighted how developers prioritize improvements and the role of Amazon's contributions, driven by their customer use cases. Chang emphasized the importance of a diverse maintainer group in decision-making and the balance between solving immediate problems and exploring creative enhancements. For more insights on MCP, tune into...
Source: The New Stack
Alex Wilhelm
2026-04-22 17:55
🌟 Exciting news in the tech world! Accenture and WaveMaker have formed a strategic partnership to help mid-market organizations modernize their applications using WaveMaker's agentic AI platform. This collaboration aims to support companies with annual revenues up to $3 billion that may lack the resources for digital transformation. The partnership focuses on delivering reliable, cost-effective AI-assisted solutions specifically for app modernization. WaveMaker’s innovative 2-pass...
Source: The New Stack
Darryl K. Taft
2026-04-22 17:49
🚀 Roo Code is transitioning from IDE-based development to cloud-based autonomous agents. The company will shut down its VS Code extension and related services on May 15. The new cloud agent, Roomote, will handle tasks across platforms like Slack and GitHub, producing outputs for developers. CEO Matt Rubens highlights the shift towards remote environments, allowing agents to complete routine tasks with minimal oversight. This evolution aims to enhance efficiency by letting developers focus on...
Source: The New Stack
Paul Sawers
2026-04-22 12:00
🚀 Exciting updates from Cloud Next '26! Google has introduced the Gemini Enterprise, which now encompasses its enterprise AI initiatives. The new Agent Platform includes tools like the Agent Studio for building agents, a simulation environment, and a marketplace for third-party agents. Key features allow agents to utilize all Google Cloud and Workspace services through the new Model Context Protocol. A web app will help teams create and manage AI agents effectively. This development marks a...
Source: The New Stack
Frederic Lardinois
2026-04-22 12:00
🚀 Google has introduced a significant change to its Tensor Processing Unit (TPU) line, unveiling two distinct chips: TPU 8t for training and TPU 8i for inference. 💡 This split aims to enhance performance as agentic workloads demand different optimizations. TPU 8t maintains its 3D torus interconnect, while TPU 8i features a new Dragonfly-inspired layout to improve communication speed. 🔍 TPU 8i also boasts increased memory capacity, designed to enhance efficiency for memory-bound tasks, crucial...
Source: The New Stack
Frederic Lardinois
2026-04-22 12:00
Google Cloud is addressing AI security challenges as attackers increasingly use AI to exploit vulnerabilities. At Next '26, the company announced new AI-powered security agents to enhance defense mechanisms across all cloud environments. 🌐🔒 These agents include a Threat Hunting agent, a Detection Engineering agent, and a Third-Party Context agent, all aimed at automating and strengthening security processes. With these tools, Google aims to reduce response times and improve detection...
Source: The New Stack
Frederic Lardinois
2026-04-22 07:00
Groundcover has expanded its AI Observability service to include native support for agentic AI systems, compatible with Google Vertex AI. This aims to provide insights into how AI systems make decisions. 🤖 The company addresses a visibility gap as organizations integrate LLMs into production, revealing challenges with traditional observability tools that were designed for deterministic systems. 🔍 Groundcover's approach allows software engineering teams to track interactions across complex...
Source: The New Stack
Adrian Bridgwater
2026-04-22 00:28
🔍 Anthropic is testing changes to its $20/month Pro plan, potentially removing access to Claude Code for new users. 📉 The company has seen increased usage, prompting adjustments to their offerings. Amol Avasare noted that existing subscribers won’t be affected and the test currently involves only 2% of new signups. 🤔 Future changes remain uncertain as they seek feedback on how to enhance user experience. #Anthropic #ClaudeCode #TechNews #AI #UserExperience
Source: The New Stack
Frederic Lardinois
2026-04-21 23:03
🚨 GitHub has paused new sign-ups for its Copilot individual plans and is tightening usage limits for current users. This decision comes as demand for AI-assisted coding increases, leading to higher compute resource needs. Joe Binder from Microsoft emphasized the need for reliability amid changing usage patterns. Since its launch in 2021, Copilot has evolved from a simple autocomplete tool to a more complex assistant that handles longer coding sessions. #GitHub #Copilot #AICoding #TechNews...
Source: The New Stack
Paul Sawers
2026-04-21 22:54
🚀 OpenAI has launched ChatGPT Images 2.0, transforming its image model into a "visual thought partner." This innovative system can reason through complex visual tasks and generate up to eight images from a single prompt. It is available to ChatGPT and Codex users, with advanced features for Plus, Pro, and Business subscribers. Images 2.0 offers two modes: Instant for quick outputs and Thinking for more deliberate reasoning, ensuring consistency in multi-frame designs. The model excels in...
Source: The New Stack
Darryl K. Taft
2026-04-21 18:45
Microsoft is exploring temporary identities to enhance security for autonomous agents in AI. During KubeCon Europe 2026, Jorge Palma, PM lead for Azure Kubernetes, highlighted a demo showcasing AI's troubleshooting capabilities. He emphasized the need for agents to have limited, temporary permissions to perform tasks safely. 🔒 The discussion also covered the cloud-to-edge continuum in AI, highlighting advancements like Microsoft’s Arc and Azure Kubernetes Service, which simplify edge AI...
Source: The New Stack
Frederic Lardinois
2026-04-21 17:15
Amazon and Anthropic have announced a significant partnership, committing over $100 billion to AWS technologies over the next decade. 🤝 Anthropic will secure up to 5 gigawatts of Trainium and Graviton cores to enhance its AI models. Additionally, the Claude Platform will be accessible on AWS, benefiting over 100,000 customers. 💻 Amazon also plans to invest up to $25 billion in Anthropic, building on a previous $8 billion investment. This collaboration highlights their ongoing relationship...
Source: The New Stack
Meredith Shubel
2026-04-21 13:00
🚨 IT uptime is a priority for organizations, with a PagerDuty study showing that 68% of companies lose over $300,000 per hour during incidents. Developer burnout is a concern, as 42% of operations leaders acknowledge its impact. Combining AI tools with human teams can alleviate this workload. However, only 37% of developers trust AI for incident response. Building this trust is essential for improving performance and reducing fatigue. #IncidentResponse #DeveloperBurnout #AI #ITUptime #TechTrends
Source: The New Stack
Mandi Walls
2026-04-21 11:00
🚀 The Eclipse Foundation has introduced the Open VSX Managed Registry, a significant step for platform engineering. This open-source alternative to Microsoft’s VS Code Marketplace serves as a central hub for developers to find and publish extensions. Open VSX supports various AI-native IDEs and platforms, ensuring reliability with a 99.95% uptime SLA. It aims to provide a secure and robust ecosystem for developers of all sizes. #EclipseFoundation #OpenSource #DeveloperTools #VSCode #OpenVSX
Source: The New Stack
Adrian Bridgwater
2026-04-21 06:00
🚀 SUSE and Nvidia have partnered to launch the SUSE AI Factory at SUSECON 2026 in Prague. This platform aims to help enterprises and governments create and manage AI solutions that meet regulatory requirements. 🛠️ Built on SUSE Rancher Prime and Nvidia AI Enterprise, it offers a standardized approach for developing AI applications across various environments. 💻 Developers can easily prototype and manage workloads, reducing setup time and operational overhead. #AI #SUSE #Nvidia #TechInnovation...
Source: The New Stack
Steven J. Vaughan-Nichols