<h2>Must know</h2>
<ul>
<li><strong>Databricks launches lakebase for AI workloads</strong>: Databricks introduced Lakebase, a new PostgreSQL-compatible database specifically engineered for AI workloads. It aims to merge operational database capabilities with their existing data lakehouse architecture, offering a serverless OLTP solution that separates compute from storage. This move, leveraging technology from Neon and Mooncake Labs, provides features like instant data branching and pgvector support, and is already generally available on AWS with Azure support in preview. <a href="https://app.daily.dev/posts/56RHAShlB" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Google embedding model deprecation causes vendor lock-in</strong>: Google deprecated its <code>text-embedding-004</code> model, rendering thousands of stored vectors unusable for a production RAG-based movie recommender. The core issue is that different embedding models create incompatible vector spaces, making migration difficult and highlighting a hidden form of vendor lock-in. The author had to re-embed their entire dataset and use a blue/green swap for their Qdrant collection, emphasizing the need to design for change when persisting embeddings at scale. <a href="https://app.daily.dev/posts/xhDI1zj4i" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Linux kernel 7.0 rc1 released with succession commentary</strong>: Linus Torvalds released Linux kernel 7.0 rc1, noting the major version bump is purely for numerical reasons, not a significant feature milestone. The release includes cemented Rust support, faster cache clearing, non-disruptive kernel updates, and updated AMD/Intel silicon support. Torvalds also mused about a future maintainer, referencing the community’s recently developed succession plan, as reported by multiple sources. <a href="https://app.daily.dev/posts/cYP8biu57" target="_blank" rel="noopener nofollow">Read more</a></li>
</ul>
<h2>Good to know</h2>
<ul>
<li><strong>OpenClaw gateway offers AI agent orchestration dashboard</strong>: Mission Control is an open-source AI agent orchestration dashboard built with Next.js 14 and TypeScript, connecting to OpenClaw Gateway via WebSocket to create tasks, plan with AI, and manage specialized agents on a Kanban board. <a href="https://app.daily.dev/posts/l0KLAjYGc" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Shuru provides local-first microvm sandbox for AI agents on macOS</strong>: Shuru is a new local-first microVM sandbox tool for macOS, using Apple’s Virtualization.framework to run lightweight Linux VMs for secure, ephemeral, and reproducible environments for AI agent workloads without Docker. <a href="https://app.daily.dev/posts/TqgsBjEoV" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>AI coding agent limitations and human developer relevance</strong>: While AI coding agents like Claude Code are highly productive, they lack the ability to devise innovative solutions when facing obstacles, a critical differentiator that preserves the relevance of human developers with creative problem-solving skills. <a href="https://app.daily.dev/posts/hB7gw6iXo" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Developer productivity hindered by non-coding tasks</strong>: A developer’s 10-second code fix for <code>flash-attention</code> took over 10 hours due to 14 steps of non-coding tasks, including fighting sandboxed test environments, CUDA version conflicts, and OOM build failures, highlighting that most programmer time is lost to abstraction layers and build systems. <a href="https://app.daily.dev/posts/OBIqfWI7x" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>VS code extension ‘fresh file explorer’ released</strong>: Fresh File Explorer is a new VS Code extension that replaces the standard file explorer to display recently modified files, integrating Git history and pending changes with features like a recency heatmap, ‘Resurrect’ for deleted files, and various grouping modes. <a href="https://app.daily.dev/posts/pQedWHBRB" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Rust concurrency performance: rwlock vs mutex for read-heavy workloads</strong>: A hands-on investigation revealed that <code>RwLock</code> performed approximately 5 times worse than <code>Mutex</code> for a read-heavy LRU tensor cache in Rust due to cache-line ping-pong from atomic increments on read locks, suggesting <code>Mutex</code> might be better for very short critical sections. <a href="https://app.daily.dev/posts/rFMs5918e" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>PostgreSQL vector search implementation alternatives</strong>: A post discusses implementing <code>pgvector</code>-like vector search functionality within PostgreSQL without relying on the <code>pgvector</code> extension, exploring DIY or alternative approaches to building vector search capabilities. <a href="https://app.daily.dev/posts/vUwhYnpqm" target="_blank" rel="noopener nofollow">Read more</a></li>
</ul>
<h2>Bullshit police</h2>
<ul>
<li><strong>Gemini 3.1 pro underperforms in coding/ui tasks vs claude opus 4.6</strong>: Despite strong benchmark scores, Google’s Gemini 3.1 Pro consistently underperformed Claude Opus 4.6 in real-world coding and UI generation tasks, with issues like tool-calling failures and incomplete feature implementation. It seems Google’s AI is still stuck in the ‘coming soon’ phase, proving that benchmarks are just numbers until they actually, you know, work. <a href="https://app.daily.dev/posts/bIAFUawfa" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Anthropic claude compiler video criticism</strong>: Anthropic released a marketing video showcasing Claude autonomously building a C compiler capable of compiling Linux and Doom, but critics are calling it ‘dishonest framing.’ Apparently, the real goal was a stress test of Claude’s autonomy, not a straightforward compiler demo. So, it’s less ‘AI builds a compiler’ and more ‘AI tries really hard to build a compiler while we film it dramatically.’ <a href="https://app.daily.dev/posts/vGYk0DbTO" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Meta patents ai for deceased user simulation</strong>: Meta was granted a patent for an AI system designed to simulate deceased users’ social media activity, including posting and chatting, using an LLM trained on their past data. Because nothing says ‘grief counseling’ like your dead grandma’s AI bot liking your vacation photos. Ethical and mental health concerns? Nah, just good old-fashioned engagement metrics. <a href="https://app.daily.dev/posts/AeW0pHCcd" target="_blank" rel="noopener nofollow">Read more</a></li>
</ul>
