<h2>Must know</h2>
<ul>
<li><strong>Google accelerates Chrome releases to fight AI browsers</strong>: Starting September 2026, Google Chrome will shift to a two-week release cycle for desktop and mobile to ship performance and security fixes faster. This move, reported by several outlets, is a direct response to ‘agentic web’ competition from OpenAI’s ChatGPT Atlas and Perplexity’s Comet. While enterprise users can stick to an eight-week cycle, Google is clearly feeling the heat from AI-native browsing experiences. <a href="https://app.daily.dev/posts/SAMWZskhL" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Apple M5 chips arrive with a side of RAM inflation</strong>: Apple’s new M5-powered MacBook Air and Pro models claim up to 4x faster AI performance, but the 14-inch Pro now starts at $2,199—a $200 hike. Analysts attribute the price jump to a global RAM shortage fueled by the AI infrastructure boom. While the hardware features Wi-Fi 7 and ‘Fusion Architecture’ for the Pro chips, Asahi Linux developers note that native Linux support for this silicon is likely years away. <a href="https://app.daily.dev/posts/FGAdSpUhd" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>North Korean StegaBin campaign targets dev environments</strong>: The Lazarus-linked group Famous Chollima has published 26 malicious npm packages that typosquat popular libraries to deploy a multi-stage infostealer. Dubbed ‘StegaBin,’ the attack uses Pastebin steganography to hide command-and-control traffic and specifically targets SSH keys, VSCode configs, and crypto wallets. Developers, particularly in Web3, should audit their dependencies for these sophisticated evasion techniques. <a href="https://app.daily.dev/posts/QDsiIqiel" target="_blank" rel="noopener nofollow">Read more</a></li>
</ul>
<h2>Good to know</h2>
<ul>
<li><strong>Google Gemini 3.1 Flash-Lite targets high-volume data processing</strong>: The new model hits 363 tokens per second at $0.25 per million input tokens, specifically optimized for scale rather than agentic orchestration. <a href="https://app.daily.dev/posts/z8znwLLdC" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Kubernetes v1.35 finally enables in-place pod resizing</strong>: You can now update CPU and memory limits on running containers without restarts, a massive win for stateful workloads like databases. <a href="https://app.daily.dev/posts/fjhms6Zq4" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Rust iterator abstractions aren’t always zero-cost</strong>: Turbopuffer engineers achieved a 60x speedup by switching to batched iterators, proving that recursive next() calls can block compiler vectorization. <a href="https://app.daily.dev/posts/U3pwVYd72" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Anthropic adds voice mode to Claude Code</strong>: The new feature allows developers to mix typing and voice commands, streaming transcribed text directly to the cursor position. <a href="https://app.daily.dev/posts/GZHhB5gpk" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Unity Asset Store to delist all Greater China publishers</strong>: Effective March 2026, thousands of assets will be removed globally due to a divergence between the main engine and the Unity China fork. <a href="https://app.daily.dev/posts/HcJsvKbOc" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Helm 4 adopts WebAssembly for plugin isolation</strong>: The new Wasm-based system replaces the legacy model, offering better portability and a 40% reduction in latency for Kubernetes extensions. <a href="https://app.daily.dev/posts/W9pY3Qv5R" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Postman pivots to AI-native Git-based workflows</strong>: A new ‘Agent Mode’ automates debugging and test writing by integrating directly with Git repositories and MCP servers. <a href="https://app.daily.dev/posts/uVCb38dOT" target="_blank" rel="noopener nofollow">Read more</a></li>
</ul>
<h2>Bullshit police</h2>
<ul>
<li><strong>The two-tier valuation shell game</strong>: AI startups like Aaru and Serval are reportedly using ‘headline’ valuations to claim unicorn status while lead VCs actually pay much less. It’s a classic vanity metric designed to trick talent and customers into thinking the company is worth $1B when the blended reality is half that. This is peak bubble behavior that will make the eventual down rounds particularly spicy. <a href="https://app.daily.dev/posts/cRCbCFvQ5" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>DevOps is dead according to the people selling AI agents</strong>: Cognition Lab claims human DevOps engineers are becoming ‘the exception’ because their Devin agent is now embedded in thousands of teams. While the tool is impressive, declaring an entire discipline dead in a marketing-heavy announcement is the kind of hyperbole we’ve come to expect. Wake us up when the AI can handle a legacy on-prem networking disaster at 3 AM without hallucinating a fix. <a href="https://app.daily.dev/posts/LfBawqtwk" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>The AI-assisted driver that couldn’t</strong>: AMD’s new DPTCi driver for gaming handhelds hit a snag when reviewers realized it was partially written by AI without disclosure. The result was a mess of ‘hallucinated tool names’ and code quality issues that required a full rewrite before it could even be considered for the kernel. It turns out ‘vibe-coding’ your way into the Linux kernel mailing list is a great way to get roasted by maintainers. <a href="https://app.daily.dev/posts/mccLSn0xk" target="_blank" rel="noopener nofollow">Read more</a></li>
</ul>
