<h2>Must know</h2>
<ul>
<li><strong>Databricks launches Lakebase for AI workloads</strong>: Databricks just dropped 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 tech from Neon and Mooncake Labs, integrates directly with the Databricks Data Intelligence Platform, allowing transactional data to be queried by Spark and Databricks SQL without the usual ETL hassle. It’s GA on AWS, with Azure in preview and GCP coming later this year. <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, succession plan hinted</strong>: The Linux kernel 7.0 rc1 is out, and Linus Torvalds says the major version bump is just because he gets confused by big numbers, not a huge feature overhaul. Still, it brings cemented Rust support, faster cache clearing, non-disruptive kernel updates, and updated AMD/Intel silicon support. Torvalds also mused about a future maintainer, hinting at the community’s recently developed succession plan. Meanwhile, the Linux ecosystem continues to evolve, with improvements like 50-80x IO uring latency fixes and PCIe 7 groundwork. <a href="https://app.daily.dev/feed-by-ids?id=cYP8biu57&amp;id=8ssuMcdSU&amp;id=JXfUDc5F4&amp;id=QdVE1yqfx&amp;id=TcabAvodN&amp;id=cQun6gD6y&amp;id=zaDpYuaxU" 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, instantly bricking thousands of stored vectors for a production RAG-based movie recommender. This highlights a nasty vendor lock-in: different embedding models create incompatible vector spaces, making migration a nightmare. The takeaway? Treat embedding models as critical infrastructure and design for change from day one, because your data is coupled to the model’s geometry, not just its API. <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 enhances AI/ML workload support</strong>: Kubernetes v1.35, codenamed ‘Timbernetes,’ is strengthening its role as the operating system for AI/ML workloads with features like workload-aware scheduling and stable in-place Pod resource resizing. <a href="https://app.daily.dev/feed-by-ids?id=mAyDHBOpw&amp;id=nytQTToPj" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>OpenAI’s Codex team uses AI for everything</strong>: OpenAI’s 40-person Codex team uses deep AI integration for onboarding, PR reviews, testing, refactors, and planning, operating with minimal hierarchy and fast iteration. <a href="https://app.daily.dev/feed-by-ids?id=8TOu6Vf1U&amp;id=4gPWtRfOI&amp;id=C3J7JVQAT&amp;id=KAYKDJGSY&amp;id=LrCmvlStu&amp;id=Lv2cFkNVA&amp;id=QwH8nwgz6&amp;id=UT5jXEM9J&amp;id=frjHTf2Ap&amp;id=iVsrjcoM3" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Intel OpenVINO 2026 boosts NPU and LLM support</strong>: Intel’s OpenVINO 2026 brings improved NPU handling and expanded support for large language models, pushing AI inference capabilities on their hardware. <a href="https://app.daily.dev/posts/IGizSZZmf" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Shuru offers local microVM sandbox for AI agents</strong>: Shuru is a new local-first microVM sandbox for macOS, designed for AI agent workloads, providing secure, ephemeral, and reproducible environments without needing Docker. <a href="https://app.daily.dev/posts/TqgsBjEoV" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Non-coding tasks kill 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 test environments, CUDA conflicts, and OOM build failures, highlighting the real productivity bottlenecks. <a href="https://app.daily.dev/posts/OBIqfWI7x" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>ChatDetector uses LLMs for RM-API misuse detection</strong>: ChatDetector, an LLM-powered system, detects resource management API misuse in software by identifying allocation APIs and pairing them to construct constraints for vulnerability detection. <a href="https://app.daily.dev/posts/UGo4Ea1t8" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Rust RwLock underperforms Mutex in read-heavy loads</strong>: A hands-on investigation revealed 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, 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>
</ul>
<h2>Bullshit police</h2>
<ul>
<li><strong>Anthropic’s Claude compiler video: pure marketing fluff</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. Because, you know, AI building a compiler from scratch is totally believable without any human intervention, right? <a href="https://app.daily.dev/posts/vGYk0DbTO" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Neiry’s bio-drones: pigeons with neural implants for surveillance</strong>: Neiry, a deeptech company, is testing ‘bio-drones’—pigeons with neural implants, solar panels, and cameras—to guide them along preset routes for infrastructure inspection and search-and-rescue. They claim “privacy compliance” through on-device processing. Because nothing says “privacy-first” like a pigeon with a brain implant and a camera, flying around and filtering out “identifiable details.” What could possibly go wrong with a flock of surveillance birds? <a href="https://app.daily.dev/posts/m7v8MA0Dh" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Fresh File Explorer: another VS Code extension you didn’t ask for</strong>: A new VS Code extension, ‘Fresh File Explorer,’ promises to replace the standard file explorer by showing recently modified files, integrating Git history, and even offering a ‘Resurrect’ feature for deleted files. It’s positioned as a lightweight alternative to GitLens. Because what developers really need is another way to look at files, especially when the existing tools are just too… functional? <a href="https://app.daily.dev/posts/pQedWHBRB" target="_blank" rel="noopener nofollow">Read more</a></li>
</ul>
