<h2>Must know</h2>
<ul>
<li><strong>Google’s embedding model deprecation causes vendor lock-in</strong>: Google deprecated its <code>text-embedding-004</code> model, rendering existing vectors unusable for a production RAG-based movie recommender. This highlights a critical issue: embedding models create a hidden vendor lock-in, as stored data becomes coupled to a specific model’s geometry. The solution involved re-embedding the entire dataset and a blue/green deployment, emphasizing the need to treat embedding models as infrastructure and design for change, as reported on February 22nd. <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, Torvalds muses on succession</strong>: Linus Torvalds released Linux kernel 7.0 rc1, a version bump he attributes to his dislike of large numbers, not a major feature milestone. This 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, according to reports from February 22nd. <a href="https://app.daily.dev/posts/cYP8biu57" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Databricks launches lakebase for AI workloads</strong>: Databricks launched Lakebase, a new PostgreSQL-compatible database designed for AI workloads, aiming to bridge operational databases with their data lakehouse architecture. This serverless OLTP database separates compute from storage, supports up to 8TB per instance, and includes Postgres 17 with pgvector. It’s currently GA on AWS, with Azure in public preview and GCP planned for later in 2026, as announced on February 22nd. <a href="https://app.daily.dev/posts/56RHAShlB" target="_blank" rel="noopener nofollow">Read more</a></li>
</ul>
<h2>Good to know</h2>
<ul>
<li><strong>Non-coding tasks are killing developer productivity</strong>: A developer’s 10-second code fix for <code>flash-attention</code> took over 10 hours due to non-coding tasks like fighting sandboxed environments, <code>CUDA</code> conflicts, and 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>Ethereum foundation commits to defi</strong>: The Ethereum Foundation is doubling down on DeFi, focusing on permissionless, censorship-resistant, privacy-first, and self-custodial financial systems, with dedicated leads for builder relationships, security, and decentralization. <a href="https://app.daily.dev/posts/gtqklfNtS" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Open source maintainers are burning out</strong>: Open source maintainers are facing burnout due to unpaid labor and lack of institutional support, with AI-generated bug reports adding to the load, prompting calls for systemic changes to recognize their contributions. <a href="https://app.daily.dev/posts/GWGLZT8P1" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Read locks are not always faster than mutexes in rust</strong>: Rust’s <code>RwLock</code> performed 5x worse than <code>Mutex</code> for a read-heavy LRU tensor cache due to cache-line ping-pong from atomic increments, 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>Shuru: a local-first microvm sandbox for AI agents on macos</strong>: Shuru is a new local-first microVM sandbox for AI agent workloads on macOS, using Apple’s Virtualization.framework to provide secure, ephemeral, and reproducible environments without Docker. <a href="https://app.daily.dev/posts/TqgsBjEoV" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>JDK 24 brings major security enhancements</strong>: JDK 24 introduces significant security enhancements, including post-quantum cryptography (ML-KEM, ML-DSA), core cryptography improvements, and the permanent removal of the Security Manager. <a href="https://app.daily.dev/posts/TLN02dr7b" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>VS code extension for recent files</strong>: The <code>Fresh File Explorer</code> VS Code extension replaces the standard file explorer to display recently modified files, integrating Git history and pending changes with a heatmap and ‘Resurrect’ feature for deleted files. <a href="https://app.daily.dev/posts/pQedWHBRB" target="_blank" rel="noopener nofollow">Read more</a></li>
</ul>
<h2>Bullshit police</h2>
<ul>
<li><strong>Neiry’s bio-drones: pigeons with neural interfaces</strong>: Neiry, a deeptech company, is testing ‘bio-drones’—pigeons with neural interfaces, solar panels, and cameras—to guide them along preset routes for infrastructure inspection. They claim ‘privacy compliance’ through on-device processing, but let’s be real, a pigeon with a camera on its head is probably not the privacy champion we’ve been waiting for. This sounds like a plot from a bad sci-fi movie, not a serious tech development. <a href="https://app.daily.dev/posts/m7v8MA0Dh" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Meta patents AI to simulate deceased users</strong>: Meta was granted a patent for an AI system that simulates a user’s social media activity, including posting and chatting, using an LLM trained on their past data, for when a user is ‘absent’ from the platform. Because nothing says ‘ethical’ and ‘mentally healthy’ like a digital zombie of your deceased loved one still liking posts. What could possibly go wrong? <a href="https://app.daily.dev/posts/AeW0pHCcd" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Anthropic’s claude compiler video: more hype than reality</strong>: Anthropic released a marketing video showcasing Claude autonomously building a C compiler capable of compiling Linux, SQLite, Lua, and Doom. Critics are calling it ‘dishonest framing,’ suggesting the real goal was a stress test of Claude’s autonomy, not a straightforward compiler-building demo. So, Claude can build a compiler, but only if you squint and ignore the marketing fluff. Classic AI hype cycle in full swing. <a href="https://app.daily.dev/posts/vGYk0DbTO" target="_blank" rel="noopener nofollow">Read more</a></li>
</ul>
