<h2>Must know</h2>
<ul>
<li><strong>Databricks launches new database for AI workloads</strong>: Databricks has unveiled Lakebase, a PostgreSQL-compatible OLTP database specifically engineered for AI workloads. This serverless offering integrates with Databricks’ data lakehouse architecture, separating compute from storage and supporting features like instant data branching and pgvector. It’s currently GA on AWS, with Azure in public preview, aiming to bridge operational databases with the lakehouse for real-time ML feature serving and AI agent memory. <a href="https://app.daily.dev/posts/56RHAShlB" 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 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, indicating long-term strategic thinking for the kernel’s future. <a href="https://app.daily.dev/posts/cYP8biu57" 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 system. This highlights a critical issue: different embedding models create incompatible vector spaces, making migration difficult and creating a hidden form of vendor lock-in. The solution involved re-embedding the entire dataset and a blue/green deployment, underscoring the need to treat embedding models as infrastructure and design for change. <a href="https://app.daily.dev/posts/xhDI1zj4i" target="_blank" rel="noopener nofollow">Read more</a></li>
</ul>
<h2>Good to know</h2>
<ul>
<li><strong>Kubernetes 1.35 strengthens AI/ML capabilities</strong>: Kubernetes v1.35, codenamed ‘Timbernetes,’ introduces features like alpha gang scheduling for distributed training and stable in-place Pod resource resize, solidifying its role as an operating system for AI/ML workloads. <a href="https://app.daily.dev/posts/mAyDHBOpw" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>OpenAI and Uber reveal infrastructure traffic control platforms</strong>: Uber’s ‘Global Rate Limiter’ handles 80 million requests per second with ‘probabilistic shedding,’ while OpenAI’s ‘Access Engine’ uses a ‘credit waterfall’ to manage traffic and ensure fair access across their systems. <a href="https://app.daily.dev/posts/OVoWP2oNj" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Rust RwLock can be slower than Mutex for read-heavy workloads</strong>: A hands-on investigation found <code>RwLock</code> performed 5x worse than <code>Mutex</code> for a read-heavy LRU cache in Rust due to cache-line ping-pong from atomic reader counter 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>Microsoft Project Silica advances glass storage</strong>: Microsoft Research’s Project Silica achieved a breakthrough in glass-based data storage, extending it to common borosilicate glass with new writing techniques and machine learning for decoding, promising data longevity of at least 10,000 years. <a href="https://app.daily.dev/posts/KqLBpGspa" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Shuru: local-first microVM sandbox for AI agents on macOS</strong>: Shuru is a new macOS tool using Apple’s Virtualization.framework to create secure, ephemeral Linux microVMs for AI agent workloads, enabling safe execution of AI-generated code without Docker. <a href="https://app.daily.dev/posts/TqgsBjEoV" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>JDK 24 brings post-quantum cryptography and security manager removal</strong>: JDK 24 introduces significant security enhancements, including post-quantum cryptography (ML-KEM, ML-DSA), SHA-3 performance gains, and the permanent removal of the Security Manager, while restricting JNI usage. <a href="https://app.daily.dev/posts/TLN02dr7b" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>New VS Code extension ‘Fresh File Explorer’ released</strong>: Fresh File Explorer is a new VS Code extension that replaces the standard file explorer, displaying recently modified files with Git history integration, color-coded recency, and a ‘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>AI coding tools: the ultimate productivity vampire</strong>: Despite all the hype and investment in AI coding tools, most organizations are seeing limited productivity gains because deployment pipelines remain the real bottleneck, with developers spending only 16% of their time actually coding. Steve Yegge even calls it an ‘energy vampire’ effect, boosting company output while burning out developers. So, while AI writes more code, you’re still stuck fighting sandboxed test environments and CUDA version conflicts for 10 hours to fix a 10-second bug. Progress! <a href="https://app.daily.dev/feed-by-ids?id=OBIqfWI7x&amp;id=Lv2cFkNVA&amp;id=QwH8nwgz6" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Meta patents AI to simulate dead users</strong>: Meta was granted a patent for an AI system that simulates deceased users’ social media activity, including posting and chatting, using an LLM trained on their past data. Because nothing says ‘grief counseling’ like a chatbot pretending to be your dead grandma. This raises ‘significant ethical and mental health concerns,’ but hey, at least Meta found a new way to keep engagement numbers up, even from beyond the grave. <a href="https://app.daily.dev/posts/AeW0pHCcd" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Anthropic’s compiler video: more marketing than magic</strong>: Anthropic released a marketing video showcasing Claude autonomously building a C compiler capable of compiling Linux and Doom. Critics, however, are calling it ‘dishonest framing,’ suggesting the real goal was a stress test of Claude’s autonomous runtime, not a straightforward compiler-building feat. So, another day, another AI company trying to make their LLM look like a wizard when it’s really just a very fast intern with a lot of pre-written scripts. <a href="https://app.daily.dev/posts/vGYk0DbTO" target="_blank" rel="noopener nofollow">Read more</a></li>
</ul>
