<h2>Must know</h2>
<ul>
<li><strong>TypeScript embraces runtime type stripping for faster server-side workflows</strong>: TypeScript is moving towards ‘no-build’ workflows for server-side development with runtime type stripping, where type annotations are removed at runtime. Node.js 22.6 now supports <code>--experimental-strip-types</code>, and TypeScript 5.8 introduced <code>erasableSyntaxOnly</code> for compatibility. This means faster server-side development by eliminating the build step, though it requires avoiding TypeScript enums and namespaces. Libraries like Zod can complement this by providing runtime type validation, as detailed in recent technical articles. <a href="https://app.daily.dev/posts/gB9bktBA8" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Vercel sandboxes get filesystem snapshots</strong>: Vercel Sandboxes now support filesystem snapshots, allowing developers to capture and reuse the complete state of a running sandbox environment. This feature streamlines workflows by eliminating repetitive setup tasks like dependency installation and build processes, enabling faster parallel testing and experimentation. The announcement from Vercel highlights improved developer productivity in CI/CD and rapid prototyping. <a href="https://app.daily.dev/posts/Uv3Zx9otG" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>New CSS techniques boost web performance</strong>: Michael Hladky highlights how CSS containment and <code>content-visibility</code> can significantly enhance web performance by reducing layout, paint, and style recalculations. These CSS-only solutions can minimize browser workload on complex pages, offering performance gains of 100ms to 500ms per page load and potentially replacing JavaScript-based virtual scrolling. Cross-browser support is now robust enough for mainstream adoption, making these accessible ‘low-hanging fruit’ for web developers, as discussed in a recent technical podcast. <a href="https://app.daily.dev/posts/ePUwFLjbL" target="_blank" rel="noopener nofollow">Read more</a></li>
</ul>
<h2>Good to know</h2>
<ul>
<li><strong>Rails 8 still the MVP king</strong>: Ruby on Rails 8 continues to be a top choice for Minimum Viable Product (MVP) development, offering built-in authentication, background jobs, caching, and WebSocket support without external dependencies, making it ideal for rapid iteration and scalability. <a href="https://app.daily.dev/posts/w2WfkMTD2" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>GitHub Copilot SDK expands agentic workflows</strong>: GitHub has released a multi-platform SDK for Python, TypeScript, Go, and .NET, allowing developers to integrate Copilot’s agentic workflows, including orchestration, planning, and programmatic file edits, into their applications. <a href="https://app.daily.dev/feed-by-ids?id=GXXnGjqMp&amp;id=GuG7K2ryZ" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Datadog unifies RUM and APM data with automatic retention filters</strong>: Datadog launched cross-product retention filters that automatically keep Application Performance Monitoring (APM) traces linked to Real User Monitoring (RUM) sessions, unifying frontend and backend observability for better troubleshooting of user experience issues. <a href="https://app.daily.dev/posts/fs5040Sq6" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Rust 1.93.0 improves musl libc and inline assembly</strong>: Rust 1.93.0 updates the bundled musl libc to version 1.2.5, improving DNS resolution for static Linux binaries, and stabilizes 23 APIs, including enhanced support for conditional inline assembly. <a href="https://app.daily.dev/feed-by-ids?id=wF6DW7Hp4&amp;id=O9g10PE7Q" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Pulumi adds <code>replacementTrigger</code> for automated resource recreation</strong>: Pulumi v3.215.0 introduces <code>replacementTrigger</code>, a new option that forces infrastructure-as-code resources to recreate automatically when specified metadata changes, useful for cycling secrets or updating external dependencies. <a href="https://app.daily.dev/posts/ay0AhT4GQ" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Doclific integrates documentation directly into code repositories</strong>: Doclific is a new tool that stores documentation directly within code repositories, aiming to prevent documentation drift by version-controlling docs alongside code and offering features like ERDs, architecture diagrams, and AI-powered generation. <a href="https://app.daily.dev/posts/vTLKU1EhC" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Meilisearch enhances enterprise search with SSO, SCIM, and Gemini embeddings</strong>: Meilisearch’s January 2026 update brings enterprise features like Single Sign-On (SSO), SCIM support, default search analytics with 20% performance improvement, dynamic sharding, and Gemini embeddings for advanced semantic search. <a href="https://app.daily.dev/posts/gySONdHAQ" target="_blank" rel="noopener nofollow">Read more</a></li>
</ul>
<h2>Bullshit police</h2>
<ul>
<li><strong>AI agents are still mostly hype for real-world tasks</strong>: Mercor’s new Apex-Agents benchmark, designed to test AI models on complex white-collar tasks, found leading models like Gemini 3 Flash and GPT-5.2 only achieved a dismal 24% accuracy. Apparently, AI agents can’t reliably replace knowledge workers in tasks requiring multi-domain reasoning and tool use. So much for the ‘AI will take all our jobs’ narrative—looks like humans still have a few years before they’re completely obsolete, at least for anything beyond basic pattern matching. <a href="https://app.daily.dev/posts/OsmrZ9UpM" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Google’s AI Mode wants to read your Gmail and Photos</strong>: Google’s new AI Mode, available to AI Pro and AI Ultra subscribers, can now access your Gmail and Google Photos to suggest vacation itineraries or shopping items. Google assures us it ‘does not train on full Gmail inboxes or photo libraries,’ only ‘specific prompts and responses.’ Because nothing says ‘privacy’ like letting an AI sift through your personal life, even if it’s ‘opt-in.’ What could possibly go wrong? <a href="https://app.daily.dev/posts/Kw6SK0Mqr" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>AI-driven code version control: because Git isn’t ‘machine-speed’ enough</strong>: Atomic Software is pushing a new version control system, based on obscure ‘patch theory,’ specifically for AI-generated code, claiming Git can’t handle ‘machine-speed’ development. Apparently, the bottleneck isn’t the AI generating questionable code, but Git’s inability to keep up with the sheer volume of it. They’ve even got a companion tool, Circuit Breaker, to restart pipelines from failure points. Because when your AI is churning out code faster than you can validate it, the obvious solution is more complex tooling, not better AI or human oversight. <a href="https://app.daily.dev/posts/xcY01cgn0" target="_blank" rel="noopener nofollow">Read more</a></li>
</ul>
