<h2>Must know</h2>
<ul>
<li><strong>Cloud security startup raises big bucks for runtime protection</strong>: Upwind Security, founded by the team behind <a href="http://Spot.io" target="_blank" rel="noopener nofollow">Spot.io</a>, secured a hefty $250 million in Series B funding, valuing the company at $1.5 billion. They’re pushing a “runtime” cloud security platform that uses an “inside-out” approach, analyzing internal signals to prioritize real-time threats, a contrast to traditional agentless scanning. This funding, led by Bessemer Venture Partners, will fuel their expansion into AI security and developer-focused misconfiguration prevention, as reported by TechCrunch. <a href="https://app.daily.dev/posts/LGgCBPA2D" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Kubernetes ingress-nginx is dead, long live Cilium</strong>: The Kubernetes ingress-nginx project is officially being archived in early 2026, meaning no more active maintenance or security updates. This affects roughly half of all cloud-native environments. Teams are urged to migrate to alternatives like Cilium Ingress for a quick switch or Cilium’s Gateway API for more advanced features, both leveraging eBPF for better performance. This move highlights the ongoing challenges of open-source project sustainability, as noted by CNCF and DEVCLASS. <a href="https://app.daily.dev/feed-by-ids?id=PptBbhnAu&amp;id=wK9Dxbori" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Google Cloud Run gets NVIDIA GPUs for serverless AI</strong>: Google Cloud Run now supports NVIDIA RTX PRO 6000 Blackwell GPUs, allowing serverless deployment of large AI models (70B+ parameters) without the hassle of cluster management. These instances scale from zero in under 5 seconds, offering 96GB vGPU memory and high bandwidth. This is a game-changer for real-time generative AI inference and model fine-tuning, making powerful AI accessible on demand in select regions, according to Google Cloud. <a href="https://app.daily.dev/posts/Vh4SzEb3w" target="_blank" rel="noopener nofollow">Read more</a></li>
</ul>
<h2>Good to know</h2>
<ul>
<li><strong>Open-source SwiftUI skill for AI coding agents</strong>: An open-source SwiftUI Agent Skill was released to help AI coding assistants build and refactor SwiftUI views, aiming to reduce technical debt by embedding best practices upfront. <a href="https://app.daily.dev/posts/LlEDXwb6h" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>AI-coded tool reviews AI-generated code locally</strong>: GitHuman, a new local-first tool for reviewing AI-generated code changes, was entirely written by an AI model and self-reviewed using GitHuman itself, shifting the code review checkpoint to the staging area. <a href="https://app.daily.dev/posts/M78k9F0lW" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Claude AI now manages Grav CMS content via MCP plugin</strong>: A developer created an open-source MCP plugin for Grav CMS, enabling Claude AI to directly manage blog content, including creation, translation, and editing, via a JSON-RPC API. <a href="https://app.daily.dev/posts/fBtDP8BjG" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Kthena orchestrates LLM inference on Kubernetes</strong>: Kthena, a new open-source sub-project of Volcano, orchestrates LLM inference on Kubernetes, addressing low GPU/NPU utilization and latency through intelligent routing and scheduling. <a href="https://app.daily.dev/posts/HONeqoore" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Wasmer boosts Python support in WebAssembly</strong>: Wasmer 7 introduces experimental async API support for Python in WebAssembly and dynamic linking in WASIX, enabling native Python modules like Numpy to run in Wasm. <a href="https://app.daily.dev/posts/fiQHrLvlC" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Rust tool sandboxes LLM agents in lightweight VMs on Macs</strong>: Vibe is a lightweight Rust tool that creates zero-configuration Linux VMs on ARM Macs for sandboxing LLM agents, offering better security and faster boot times than containers. <a href="https://app.daily.dev/posts/XJySsKWNc" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Joly UI offers accessible React components</strong>: Joly UI offers a collection of accessible React components built on shadcn/ui and Radix UI, fully typed with TypeScript and themeable with Tailwind CSS, for quick UI development. <a href="https://app.daily.dev/posts/EG6uDXQBi" target="_blank" rel="noopener nofollow">Read more</a></li>
</ul>
<h2>Bullshit police</h2>
<ul>
<li><strong>Tesla robotaxis: crashing more, disclosing less</strong>: Tesla’s robotaxi fleet in Austin reportedly has a crash rate three times worse than human drivers for all incidents, and nine times worse for police-reported crashes, despite having safety monitors in every vehicle. Meanwhile, Tesla continues to redact all crash narratives in its NHTSA reports, unlike competitors like Waymo who operate truly driverless vehicles with a better safety record. Transparency? Who needs it when you have a cult following? <a href="https://app.daily.dev/posts/M8WkrOIKM" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Microsoft’s AI overload: a masterclass in listening to users (eventually)</strong>: Microsoft is apparently “walking back” its aggressive AI integration in Windows 11 after significant user backlash. They’re “reviewing” Copilot integrations in apps like Notepad and Paint, and even considering “evolving or renaming” Windows Recall. So, after shoving AI down everyone’s throats, they’re now pretending to listen. What a shocker that users don’t want half-baked, privacy-invasive features forced upon them. <a href="https://app.daily.dev/posts/mR53bImT0" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>AI expectations: the new management tool for developer misery</strong>: A survey of Norwegian developers reveals growing workplace dissatisfaction, with 28% reporting increased workload due to “AI expectations” from management. Apparently, leadership expects AI to magically solve all problems, leading to unrealistic demands and poor technical leadership. It seems the only thing AI is consistently generating is developer burnout and a healthy dose of skepticism. <a href="https://app.daily.dev/posts/EiYsnniH2" target="_blank" rel="noopener nofollow">Read more</a></li>
</ul>
