<h2>Must know</h2>
<ul>
<li><strong>Orbital compute reshapes the AI race</strong>: The AI race is shifting from algorithms to infrastructure control, with the SpaceX-xAI merger creating a vertically integrated powerhouse. They now command rocket manufacturing, a massive Starlink constellation, the Colossus supercomputer, and plans for an orbital data center. This move is driven by terrestrial bottlenecks like electricity grids and permitting, making orbital compute with continuous solar power and vacuum cooling a strategic advantage, as highlighted by an unknown source. Other players like Amazon and the EU are trying to catch up, but lack the same level of integration. <a href="https://app.daily.dev/posts/QqzFDNqqS" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Gemini 3.1 Pro: benchmark king, usability flop</strong>: Google’s Gemini 3.1 Pro, despite boasting top benchmark scores and a low cost, is reportedly a letdown in real-world coding and UI generation tasks. Multiple sources, including articles titled “Gemini 3.1 Pro is the smartest model ever made” and “Gemini 3.1 PRO…pure disappointment,” point to issues like tool-calling failures, incomplete features, and poor environment detection. While it excels on benchmarks, its practical usability for complex agentic workflows is significantly hampered, especially when compared to models like Claude Opus 4.6. <a href="https://app.daily.dev/feed-by-ids?id=bIAFUawfa&amp;id=mGa3eARS3" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>AI agents are creating a security debt crisis</strong>: AI agents are introducing a ‘security debt crisis’ by prioritizing speed over secure practices, leading to vulnerabilities like hardcoded API keys and public access policies, according to Columbia University research. Stripe’s “Minions” autonomously generate over 1,000 pull requests weekly, showcasing the scale of AI-driven code. Mitigations include spec-driven prompting with OWASP Top 10 and CI/CD scanners, but the core issue remains: AI agents are fast, but not necessarily secure. <a href="https://app.daily.dev/feed-by-ids?id=DD6LZnw0Q&amp;id=Et2d8uHu3&amp;id=LUO8dPKde&amp;id=RH1TYGIET&amp;id=UGo4Ea1t8&amp;id=bDJrJKMJX&amp;id=em7afQ5sm&amp;id=l0KLAjYGc&amp;id=ohhVPexbM&amp;id=vXBWwcwBk" target="_blank" rel="noopener nofollow">Read more</a></li>
</ul>
<h2>Good to know</h2>
<ul>
<li><strong>Open-source multi-cluster management for Proxmox VE</strong>: PegaProx is an open-source, web-based multi-cluster management dashboard for Proxmox VE, offering unified control, live metrics, and advanced features like load balancing and LDAP/OIDC authentication. <a href="https://app.daily.dev/posts/zaDpYuaxU" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Run Llama 3.1 70B on a single RTX 3090</strong>: NTransformer, an open-source C++/CUDA LLM inference engine, claims to run Llama 3.1 70B on a single RTX 3090 by streaming layers via PCIe and using a 3-tier adaptive caching system. <a href="https://app.daily.dev/feed-by-ids?id=Ahx4zgQks&amp;id=WoJMuyqAQ" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>LangSmith Agent Builder uses filesystem-based memory</strong>: LangSmith Agent Builder uses a filesystem-based memory system, storing memory as files in Postgres but exposing it as a virtual filesystem to give task-specific agents persistent knowledge across sessions. <a href="https://app.daily.dev/posts/9xLrRFK2b" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Elixir language server Expert reaches release candidate</strong>: The official Elixir language server, Expert, has reached its first release candidate, merging Lexical and Next LS to provide compiler diagnostics, completions, and monorepo support. <a href="https://app.daily.dev/posts/ItEtgwo7d" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Go’s FFI now faster than GDScript and Rust</strong>: Go’s FFI performance for Godot via GDExtension now surpasses GDScript and even ‘safe’ Rust, thanks to call batching and Go 1.26’s cgo overhead reduction. <a href="https://app.daily.dev/posts/j8AoMJYTv" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Anthropic and Figma enable code-to-design workflow</strong>: Anthropic and Figma collaborated on a new workflow that allows Claude Code to convert production code into editable Figma designs, facilitating design collaboration and round-tripping changes back to code. <a href="https://app.daily.dev/posts/aCaE5xJ9z" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>KittenTTS: ultra-compact text-to-speech for edge devices</strong>: KittenTTS by Kitten ML offers ultra-compact text-to-speech models (down to 15M parameters) optimized for edge and browser deployment, running client-side on CPU with Apache 2.0 license. <a href="https://app.daily.dev/posts/aXA1rlMhE" target="_blank" rel="noopener nofollow">Read more</a></li>
</ul>
<h2>Bullshit police</h2>
<ul>
<li><strong><a href="http://Lobste.rs" target="_blank" rel="noopener nofollow">Lobste.rs</a> learns SQLite isn’t a MariaDB replacement the hard way</strong>: <a href="http://Lobste.rs" target="_blank" rel="noopener nofollow">Lobste.rs</a>, a popular tech news aggregator, attempted to migrate from MariaDB to SQLite and quickly reverted due to “severe performance degradation.” Apparently, the team had considered PostgreSQL but settled on SQLite. It’s almost like a database designed for single-user local storage isn’t ideal for a high-traffic web application. Who could have seen that coming? <a href="https://app.daily.dev/posts/RLQeRSeS0" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Developer discovers fixing tools is faster than working around them</strong>: A developer shared a personal anecdote about working around a broken debugger instead of fixing it, only to realize a one-line config change would have saved hours. The takeaway: “fix your tools.” Groundbreaking. It’s almost like we’re paid to solve problems, not just blindly bash our heads against them. Next, they’ll tell us to read the manual. <a href="https://app.daily.dev/posts/xN5zIaynw" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Senior engineer discovers the revolutionary concept of asking questions</strong>: A senior engineer shared their profound mentorship philosophy: when disagreeing with a teammate, start with questions, not answers. They even provided a real-world example of challenging a “seemingly obvious” round-robin approach. Because, you know, asking questions and fostering critical thinking is totally new in engineering. What’s next, suggesting we communicate effectively? <a href="https://app.daily.dev/posts/zdeI6Uw9R" target="_blank" rel="noopener nofollow">Read more</a></li>
</ul>
