<h2>Must know</h2>
<ul>
<li><strong>Google rolls out Gemini 3.1 Pro preview</strong>: Google just dropped Gemini 3.1 Pro into public preview, touting it as their latest agentic coding and state-of-the-art reasoning model. It scored 77.1% on the ARC-AGI-2 benchmark, a significant jump from its predecessor, and leads most coding benchmarks. This model supports multimodal inputs and offers a 1 million token context window, intensifying the AI model race with competitors like Anthropic and OpenAI, as reported by multiple sources. <a href="https://app.daily.dev/posts/PjE6Xo1XG" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>MySQL governance under scrutiny</strong>: Over 248 database professionals, including engineers from Percona and MariaDB, have signed an open letter urging Oracle to transfer MySQL governance to an independent non-profit foundation. They cite declining commit activity, a shrinking contributor pool, and a lack of AI-era features as key concerns. The community believes this move is crucial to revitalize MySQL development and ensure its future relevance, drawing parallels to PostgreSQL’s community-led success. <a href="https://app.daily.dev/posts/GY7vcPtOu" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>AI agent adoption and impact in software development</strong>: AI coding tools are rapidly shifting developer roles from writing code to reviewing and orchestrating AI agents, with 93% of developers now using AI assistants. While some report significant productivity gains, concerns are rising about AI-generated code introducing subtle production bugs and the ‘AI slop problem’ in open source, where low-quality AI contributions create extra review burden for maintainers. This highlights the need for robust engineering maturity and new developer skills in context fluency for effective agentic coding. <a href="https://app.daily.dev/feed-by-ids?id=037a5YIjx&amp;id=0Q86tGAco&amp;id=0tYkAozTe&amp;id=1v9gMZMqJ&amp;id=2LocixQWJ&amp;id=2WWQ65Wm6&amp;id=2Xamq3Lpe&amp;id=3Kdb5qCcA&amp;id=4CjaWv3pP&amp;id=6KUpHm9AN" target="_blank" rel="noopener nofollow">Read more</a></li>
</ul>
<h2>Good to know</h2>
<ul>
<li><strong>Vitest 3.x crushes Jest 30 in performance benchmarks</strong>: Vitest 3.x shows 5.6x faster cold starts and 28x faster watch mode re-runs compared to Jest 30 in a 50,000-test monorepo, along with 57% less peak memory usage. <a href="https://app.daily.dev/feed-by-ids?id=60rKeST9w&amp;id=9RmovJfdu&amp;id=GsL0A9M1P&amp;id=T3G4VUzdk&amp;id=yRN3HZsMy" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>GitOps best practices for Kubernetes</strong>: A guide to 12 GitOps best practices for Kubernetes, emphasizing Git as the single source of truth, directory-based environment separation, and using commit SHAs for image tagging. <a href="https://app.daily.dev/posts/KF2N6PqEu" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Node.js memory usage halved with node-caged</strong>: Cloudflare, Igalia, and the Node project collaborated on <code>node-caged</code>, a Docker image with V8 pointer compression that can halve Node.js memory usage in production. <a href="https://app.daily.dev/posts/Mm7b9KCfF" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Workload identity federation now GA for Tailscale</strong>: Tailscale’s workload identity federation is now generally available, allowing CI/CD pipelines and cloud workloads to authenticate using short-lived OIDC tokens instead of static credentials. <a href="https://app.daily.dev/posts/ekx9IIs1v" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Spanner now supports native CQL APIs</strong>: Google Cloud’s Spanner now offers general availability for native CQL (Cassandra Query Language) API support, enabling Cassandra application migration with minimal code changes and offering global ACID transactions. <a href="https://app.daily.dev/posts/CKaLROkeW" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>DNF replaces YUM in 2026</strong>: YUM is officially deprecated in 2026, with DNF (Dandified YUM) becoming its modern replacement on RHEL, Fedora, and Rocky Linux, offering smarter dependency resolution and parallel downloads. <a href="https://app.daily.dev/posts/GT9dzQz9f" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>KFC architecture blueprint for real-time data</strong>: The KFC Architecture Blueprint combines Kafka, Flink, and ClickHouse for end-to-end real-time data processing, suitable for fraud detection, clickstream analytics, and IoT telemetry. <a href="https://app.daily.dev/posts/pAQO9MXVw" target="_blank" rel="noopener nofollow">Read more</a></li>
</ul>
<h2>Bullshit police</h2>
<ul>
<li><strong>Claude Pro integrates with Microsoft PowerPoint</strong>: Oh, joy! Claude Pro now integrates with Microsoft PowerPoint, allowing users to ‘pull context from their daily tools directly into presentations.’ Because what we really needed was another AI to help us make even more slides, probably filled with even more AI-generated corporate jargon. The promotional material claims Claude Pro is ‘getting bigger every week,’ which sounds suspiciously like a growth spurt fueled by hot air and marketing budgets. <a href="https://app.daily.dev/posts/oLQ4V0bPw" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Trajectory Explorer tool announcement</strong>: Someone announced ‘Trajectory Explorer,’ a new tool for ‘navigating AI agent traces’ that promises a ‘sane way’ to inspect AI agent decisions. The announcement, made via retweeted posts, uses phrases like ‘we’re excited’ and ‘the first sane way.’ Because apparently, before this, everyone was just blindly trusting their AI agents to make decisions in an insane, uninspectable void. Color me skeptical. <a href="https://app.daily.dev/posts/AzPQg0qJn" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Don’t use Next.js if you’re building for the long term</strong>: A brave soul dared to suggest that Next.js might not be the best choice for long-term projects, citing ‘upgrade churn, tight coupling to Vercel’s ecosystem, and implicit behavior traps.’ Imagine that, a framework that ties you to a specific vendor and constantly changes under your feet might not be ideal for something you want to last. Who would’ve thought? It’s almost like chasing the latest shiny object has consequences. <a href="https://app.daily.dev/posts/0vMYzSZd0" target="_blank" rel="noopener nofollow">Read more</a></li>
</ul>
