<h2>Must know</h2>
<ul>
<li><strong>Kubernetes is now the standard for enterprise AI</strong>: The 2025 CNCF Annual Cloud Native Survey reveals that Kubernetes production usage has hit 82%, solidifying its role as the standard infrastructure for enterprise workloads, especially AI. For AI workloads, 66% of adopters use Kubernetes for inference. Interestingly, the survey highlights that cultural resistance, not technical complexity, is now the primary barrier to adoption, according to the CNCF. This shift indicates that while the tech is mature, organizational dynamics are the new bottleneck. <a href="https://app.daily.dev/posts/Lwsgn3lZ5" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>AI is accelerating exploit and malware development</strong>: New research indicates that AI, specifically LLMs like Opus 4.5 and GPT-5.2, can generate sophisticated exploits for zero-day vulnerabilities, even bypassing modern security mitigations. One experiment showed GPT-5.2 solving a complex scenario in just three hours for about $50 in compute. Separately, Check Point Research identified VoidLink, the first known AI-generated malware suite, developed by a single person using AI agents in under a week—a task estimated to take multiple human teams 30 weeks. This suggests AI is industrializing offensive cybersecurity, shifting the bottleneck from human expertise to LLM throughput. <a href="https://app.daily.dev/feed-by-ids?id=WkT9zuzY1&amp;id=nLTC7GTyu" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Microsoft mandates Intune security updates for mobile apps</strong>: Microsoft is enforcing mandatory Intune Mobile Application Management security updates starting January 19, 2026. Enterprises must update their iOS and Android apps to the latest Intune SDK versions, or risk having them blocked from accessing corporate resources like Outlook and Teams. This update introduces new security controls such as PIN requirements and biometric authentication, and there’s no grace period, meaning non-compliant apps will be immediately blocked. <a href="https://app.daily.dev/posts/PNEoUht10" target="_blank" rel="noopener nofollow">Read more</a></li>
</ul>
<h2>Good to know</h2>
<ul>
<li><strong>AI coding agents show velocity gains but need guardrails</strong>: AI coding agents like Claude Code can deliver significant velocity gains (10–100x) for well-defined tasks, but struggle with ambiguous requirements, often leading to code repetition and omitted tests, necessitating careful human oversight and strong guardrails like automated quality checks and peer reviews. <a href="https://app.daily.dev/feed-by-ids?id=EqMZfdLoq&amp;id=GPLZOgOgP&amp;id=J24eplQdr&amp;id=cSbwL1BS5" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>New local AI coding model GLM-4.7 Flash outperforms competitors</strong>: GLM-4.7 Flash, a 30B-A3B mixture-of-experts language model, is claiming top performance in local AI coding benchmarks like SWBench and Tau Squared Bench, significantly outperforming competitors like Qwen 3-30B-A3B, particularly in tool calling and agentic tasks, and is available for self-hosting with an MIT license. <a href="https://app.daily.dev/posts/eWdzokgG0" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Taiwan and US sign AI and semiconductor supply chain pact</strong>: Taiwan and the United States have finalized a strategic trade agreement focused on securing AI and semiconductor supply chains, with Taiwan committing $250 billion in investments and credit guarantees for US-based production, and Taiwanese chip makers like TSMC receiving lowered tariffs and duty-free export status for some items. <a href="https://app.daily.dev/posts/suaVSewS2" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Next.js 16 boosts performance and AI debugging</strong>: Next.js 16 stabilizes Turbopack as the default bundler for faster builds and hot reloading, introduces explicit Cache Components, and integrates AI-assisted debugging via Model Context Protocol (MCP) in DevTools, allowing AI agents to suggest fixes from unified logs. <a href="https://app.daily.dev/posts/91xiEmcVe" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Elysia.js offers type-safe APIs for Bun</strong>: Elysia.js is a new TypeScript-first API framework for Bun that simplifies API route organization, provides end-to-end type safety without code generation via its Eden companion library, and uses TypeBox for inline schema validation and TypeScript type generation. <a href="https://app.daily.dev/posts/pGvKxqMl2" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>New tool streamlines cloud-native testing on Kubernetes</strong>: MetalBear released mirrord for CI, a new tool that allows CI tests to run directly on shared Kubernetes clusters, eliminating the need for ephemeral environments and simulating production conditions, which MetalBear claims saves 20-30 minutes per CI run and reduces cloud costs. <a href="https://app.daily.dev/posts/s2plqXXTq" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Anthropic fixes critical Git server vulnerabilities</strong>: Anthropic patched three critical vulnerabilities in its Git MCP server (CVE-2025-68145, CVE-2025-68144, CVE-2025-68143) that could have allowed remote code execution via prompt injection, highlighting new security risks in AI-integrated systems. <a href="https://app.daily.dev/posts/U6A8HrCeq" target="_blank" rel="noopener nofollow">Read more</a></li>
</ul>
<h2>Bullshit police</h2>
<ul>
<li><strong>AI coding tools: the new executive fantasy</strong>: Apparently, ‘vibecoding’ is the new hotness, with Emergent raising $70 million for its AI-powered app builder. An opinion piece even suggests executives should be ‘vibecoding’ to build internal tools, bypassing those pesky engineering teams. Because nothing says ‘efficient’ like a CEO trying to code a metric-visualization tool with an AI, right? What could possibly go wrong? <a href="https://app.daily.dev/feed-by-ids?id=AllrxKqHW&amp;id=McILqauKu" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>AI will make you obsolete, unless you pay for my course</strong>: The usual suspects are at it again, pushing the narrative that AI will eliminate software engineering jobs, conveniently ignoring the fact that 56% of CEOs see no revenue increase or cost decrease from AI spending. Meanwhile, some YouTube creators are lamenting declining viewership for dev content due to AI adoption, while others are still trying to sell you on the ‘best programming languages to learn in 2026.’ It’s almost like they want you to panic and buy their stuff. <a href="https://app.daily.dev/feed-by-ids?id=RJIgj9I8E&amp;id=Y2d0smXfY&amp;id=hysT7bS89&amp;id=codinginpublic" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>The never-ending AI hype cycle</strong>: Another day, another AI model release, another round of exaggerated claims and unrealistic expectations. Industry observers note this cycle repeats with every major AI launch: initial euphoria, followed by disappointment as limitations emerge. Social media amplifies the hype, pushing predictions of job displacement, while analysts warn that discussions often ignore drawbacks. It’s almost as if some people benefit from this constant cycle of over-promising and under-delivering. <a href="https://app.daily.dev/feed-by-ids?id=j0xF0WiVr&amp;id=yCkgCoa6x" target="_blank" rel="noopener nofollow">Read more</a></li>
</ul>
