<h2>Must know</h2>
<ul>
<li><strong>Databricks launches lakebase for AI workloads</strong>: Databricks just dropped Lakebase, a new PostgreSQL-compatible database specifically for AI workloads. It’s designed to merge operational database capabilities with their lakehouse architecture, offering serverless OLTP, compute-storage separation, and features like instant data branching and pgvector support. This is a big move for AI infrastructure, especially with its GA on AWS and Azure preview, as reported by Databricks. <a href="https://app.daily.dev/posts/56RHAShlB" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>LangChain agent builder gets persistent memory</strong>: LangChain’s Agent Builder now uses a filesystem-based memory system, giving agents persistent, evolving knowledge across sessions. This virtual filesystem, stored in Postgres, maps to COALA’s memory taxonomy, allowing agents to update their own memory in real-time. They’re even planning human-in-the-loop approval to prevent prompt injection, which is a smart move for AI agent security. <a href="https://app.daily.dev/posts/9xLrRFK2b" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Google embedding model deprecation causes RAG system chaos</strong>: Google deprecated its <code>text-embedding-004</code> model, rendering thousands of stored vectors useless for a production RAG system. This highlights a critical vendor lock-in issue: different embedding models create incompatible vector spaces, making migration a nightmare. The takeaway is to treat embedding models as infrastructure and design for change if you’re persisting embeddings at scale, as one developer learned the hard way. <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>OpenClaw AI agent use cases and limitations</strong>: OpenClaw, an autonomous AI agent tool, offers use cases like automated dependency maintenance, daily tech research briefings, and cloud cost monitoring, but requires pre-configured permissions or a <code>dangerously-skip-permissions</code> flag for Claude Code integration. <a href="https://app.daily.dev/posts/WCj2dFu7Q" 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 macOS, designed for AI agent workloads, providing secure, ephemeral, and reproducible environments using Apple’s Virtualization.framework without needing Docker. <a href="https://app.daily.dev/posts/TqgsBjEoV" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>WebMCP: browser-native execution for AI agents</strong>: Google’s WebMCP is a browser-native API allowing websites to expose structured JavaScript tools for AI agents to invoke directly, reducing token overhead and improving determinism compared to traditional browser automation. <a href="https://app.daily.dev/posts/61GmU0Tcy" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Linux kernel 7.0 rc1 released with rust support</strong>: Linux kernel 7.0 rc1 is out, featuring stable Rust support, faster cache clearing, non-disruptive kernel updates, and performance improvements for RISC-V and LoongArch. <a href="https://app.daily.dev/posts/cYP8biu57" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Fresh file explorer: a new VS Code extension</strong>: A new VS Code extension, ‘Fresh File Explorer,’ replaces the standard file explorer to display recently modified files, integrating Git history and offering features like a recency heatmap and one-click restore for deleted files. <a href="https://app.daily.dev/posts/pQedWHBRB" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Developer productivity lost to non-coding tasks</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 version conflicts, and OOM build failures, highlighting how environmental setup overshadows actual coding. <a href="https://app.daily.dev/posts/OBIqfWI7x" 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 and demonstrating data longevity of at least 10,000 years, with research findings now public. <a href="https://app.daily.dev/posts/KqLBpGspa" target="_blank" rel="noopener nofollow">Read more</a></li>
</ul>
<h2>Bullshit police</h2>
<ul>
<li><strong>Anthropic’s claude compiler video: marketing or magic?</strong>: Anthropic released a marketing video showing Claude autonomously building a C compiler capable of compiling Linux and Doom. Critics, however, are calling it “dishonest framing,” suggesting the real goal was merely a stress test of Claude’s ability to run autonomously for an extended period. So, it seems Claude can build a compiler, but Anthropic might be compiling a bit of marketing fluff. <a href="https://app.daily.dev/posts/vGYk0DbTO" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Gemini 3.1 pro: benchmark king, real-world flop?</strong>: Google’s Gemini 3.1 Pro, despite its impressive benchmark scores, apparently underperformed Claude Opus 4.6 in real-world coding and UI generation tasks. One “hands-on comparison” called it “pure disappointment,” citing tool-calling failures and incomplete feature implementation. Guess those benchmarks don’t always translate to actual usefulness, who would’ve thought? <a href="https://app.daily.dev/posts/bIAFUawfa" 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 that simulates deceased users’ social media activity, including posting and chatting, using an LLM trained on their past data. The patent describes its use when a user is ‘absent’ from the platform. Because what we all need is our dead relatives’ AI ghosts liking our vacation photos. Ethical and mental health concerns? Nah, just more engagement! <a href="https://app.daily.dev/posts/AeW0pHCcd" target="_blank" rel="noopener nofollow">Read more</a></li>
</ul>
