<h2>Must know</h2>
<ul>
<li><strong>AI coding tools are changing everything, but not how you think</strong>: AI coding assistants are now generating a significant portion of code, with some developers shipping mostly AI-written code. This shift is moving the bottleneck from writing code to verifying it, as AI-generated code often contains more logic errors and security flaws. The focus for engineers is moving towards code review, evaluating AI output, and managing increased pull requests, rather than writing code from scratch. This means new infrastructure and verification strategies are needed to handle the volume and ensure quality, as highlighted by multiple industry reports and developer accounts. <a href="https://app.daily.dev/feed-by-ids?id=T7lz8DVGQ&amp;id=gahr7f5EZ&amp;id=ggxtFI8fz&amp;id=imPFZmJZw&amp;id=jOUUi1fNS&amp;id=zZPpIk0MP&amp;id=1lxfdifYi&amp;id=ueD6Vp8hW&amp;id=zANf1p7HZ&amp;id=DuxmOezZE" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>AI agents are here, but so are the security nightmares</strong>: Agentic AI systems, capable of autonomous computer use and coding, are bringing a new wave of security vulnerabilities. Researchers at Radware discovered ‘ZombieAgent,’ a zero-click indirect prompt injection vulnerability in OpenAI’s agent platforms, allowing persistent, covert actions. IBM’s AI coding agent ‘Bob’ also has critical flaws enabling malware execution via prompt injection. These incidents highlight that while AI agents offer immense potential, they also introduce new attack surfaces that traditional security tools struggle to detect, as reported by Radware and other security researchers. <a href="https://app.daily.dev/feed-by-ids?id=j249LcDmA&amp;id=CP910mpyj&amp;id=EDYtMcTfL&amp;id=60HE5QUGv&amp;id=CuSSpmwf6&amp;id=GTsss02X9&amp;id=hOkZZEdlQ&amp;id=RiaoDNM5O&amp;id=TVFNFf82Y" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Global memory chip shortage drives up prices for everything</strong>: A severe global shortage of DRAM and NAND memory chips is causing prices to skyrocket, impacting everything from firewalls to consumer electronics. Samsung projects a threefold increase in Q4 2025 operating profit due to surging memory prices, with DRAM contract prices up 313% annually. This shortage, driven by massive AI data center demand for high-bandwidth memory, means PC manufacturers are already reducing RAM in budget systems, and experts predict elevated prices will persist for years, according to Samsung and industry analysts. <a href="https://app.daily.dev/feed-by-ids?id=RV04f4xGD&amp;id=YUGmJe9Ca&amp;id=fj4JypEc8" target="_blank" rel="noopener nofollow">Read more</a></li>
</ul>
<h2>Good to know</h2>
<ul>
<li><strong>Google Gemini 3 expands multimodal AI and developer tools</strong>: Google’s Gemini 3 platform now processes text, images, video, audio, and PDFs within a unified multimodal model, simplifying development and reducing the need for separate AI pipelines, with a 1M+ token context window and Deep Think mode for complex reasoning. <a href="https://app.daily.dev/feed-by-ids?id=UyBvDFk8k&amp;id=0CxKgm4eV&amp;id=NYhUyiiE6&amp;id=wJjYqm8Xj&amp;id=IYZKv2e0G&amp;id=laUABpjHh&amp;id=vZbhC083l&amp;id=QaeYUARam&amp;id=nLPqowIJN&amp;id=qjYQj2ENJ" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>OpenAI and Anthropic launch competing AI health platforms</strong>: OpenAI launched ChatGPT Health, allowing users to upload medical records for insights, while Anthropic countered with Claude for Healthcare, focusing on HIPAA-compliant administrative workflows for providers and payers. <a href="https://app.daily.dev/feed-by-ids?id=52NiUSgqi&amp;id=WF5GxdsgD&amp;id=FGI4Oyps1&amp;id=G2f0bA0Zl" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Microsoft open-sources XAML Studio, acquires Osmos</strong>: Microsoft open-sourced XAML Studio, a rapid prototyping tool for WinUI developers, and acquired Osmos, an AI data engineering startup, to enhance its Fabric data platform with AI automation. <a href="https://app.daily.dev/feed-by-ids?id=8ioHmu8f0&amp;id=Tm2Js37ee&amp;id=dRdzpOrJb&amp;id=eRi1VQ5SP" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>New open-source AI coding tools emerge</strong>: Void, a new open-source code editor based on VS Code, is gaining traction, offering self-managed AI API keys and private LLM hosting, while Kilo Code and Vibe Kanban provide open-source platforms for managing multiple AI coding agents. <a href="https://app.daily.dev/feed-by-ids?id=nD0XCJXOA&amp;id=QJSZXW8qJ&amp;id=VSCocWaDE&amp;id=WtQOA9MEd&amp;id=f8luCQfKk&amp;id=t9TmWNeD6&amp;id=0jkLgIBom" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Linux kernel and QEMU advance, AMD standardizes server firmware</strong>: The Linux kernel is removing the legacy mount API and adding RISC-V Spectre V1 mitigations, QEMU 11.0 drops 32-bit host CPU support, and AMD is standardizing firmware for next-gen EPYC server SoCs to improve interoperability. <a href="https://app.daily.dev/feed-by-ids?id=50YPqvTFE&amp;id=BqO1pYoWM&amp;id=FlCRi4J5x&amp;id=i5xvLJ6Yb&amp;id=iJHSBJrAw&amp;id=ijNx4P4vv&amp;id=eeFdKxuyc" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>New tools for AI agent memory and observability</strong>: NevaMind-AI released MemU, an open-source multimodal memory infrastructure for LLM agents, while Elasticsearch 9.2 introduces an AI Agent Builder and Streams for AI-powered log summarization, and Gemini CLI adds monitoring dashboards with OpenTelemetry. <a href="https://app.daily.dev/feed-by-ids?id=JWlybrZkg&amp;id=yhx6bvEhz&amp;id=yvVll2Ohq" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Go ecosystem analysis reveals top dependencies</strong>: An analysis of the Go ecosystem shows ‘<a href="http://github.com/stretchr/testify" target="_blank" rel="noopener nofollow">github.com/stretchr/testify</a>’ as the most popular dependency, with insights into dependency patterns and the continued use of deprecated libraries, all processed using Neo4j. <a href="https://app.daily.dev/posts/heMEIyzLQ" target="_blank" rel="noopener nofollow">Read more</a></li>
</ul>
<h2>Bullshit police</h2>
<ul>
<li><strong>AI is ‘killing’ open source, or maybe just bad business models?</strong>: Tailwind CSS laid off 75% of its engineering team after an 80% revenue drop, blaming AI models for scraping their docs and reducing website traffic. Apparently, developers using AI to generate code means they don’t need to visit the official docs, thus bypassing the ‘funnel’ for paid products. So, AI is ‘killing’ open source, or maybe it’s just exposing the fragility of business models that rely on documentation traffic instead of, you know, actual product value or sustainable licensing. As one author put it, ‘AI Killed One Open Source Business Model,’ not open source itself. <a href="https://app.daily.dev/feed-by-ids?id=LLQ1zfubn&amp;id=f3fYMXuKZ&amp;id=hbFeNQ18o&amp;id=d57ILPjBN&amp;id=re8TItYuD&amp;id=2a8HRxzCb&amp;id=u4S1Vnazf&amp;id=Ml47Y2Hd6&amp;id=mpCeVdjI4&amp;id=RuMhQtj6W" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Anthropic’s ‘walled garden’ for Claude Code: innovation or cash grab?</strong>: Anthropic decided to restrict Claude Code and Claude Pro subscriptions from third-party tools, forcing developers to use their official interface or pay for higher-cost API plans. Developers are, predictably, calling foul, accusing Anthropic of anti-competitive practices and building a ‘walled garden’ akin to Apple’s. Meanwhile, OpenAI is strategically supporting third-party integrations with ChatGPT. It seems Anthropic is prioritizing ecosystem control over developer freedom, which is a bold move when your competitors are literally doing the opposite, as noted by developer community blogs. <a href="https://app.daily.dev/feed-by-ids?id=PQFa1I0wt&amp;id=jHVIJy7Cw&amp;id=dRdzpOrJb&amp;id=eRi1VQ5SP" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>The ‘AI Detox’ for engineers: because AI is making us dumber, apparently</strong>: Apparently, AI code generation tools are causing ‘dead zones’ of 5-30 seconds, leading engineers to context-switch to Slack or email, thus destroying ‘deep work.’ The proposed solution? An ‘AI Detox’ protocol: remove hands from the keyboard, close distracting apps, and ‘actively engage with the AI’s thought process.’ It’s almost as if we’re being told to meditate while our AI overlords churn out code. Maybe the real problem isn’t the AI, but the expectation that we should be glued to our screens 24/7, or perhaps the tools themselves are just not good enough yet, as suggested by articles on engineer productivity. <a href="https://app.daily.dev/feed-by-ids?id=eC9HsHF69&amp;id=infazu71Y&amp;id=nVFL2FYYR&amp;id=uTQFFroPQ" target="_blank" rel="noopener nofollow">Read more</a></li>
</ul>
