<h2>Must know</h2>
<ul>
<li><strong>Google’s Gemini 3.1 Pro is here</strong>: Google just dropped Gemini 3.1 Pro, their latest agentic and coding model, now in public preview. It boasts a massive 1 million token context window and scored 77.1% on the ARC-AGI-2 benchmark, a huge leap from its predecessor. This model is available across various platforms, including GitHub Copilot, Google AI Studio, and Vertex AI, and is priced competitively at $2 per million input tokens. <a href="https://app.daily.dev/posts/PjE6Xo1XG" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Anthropic tightens Claude code access</strong>: Anthropic updated its Claude Code documentation, restricting OAuth token usage from Pro/Max accounts in third-party tools, causing a stir among developers. While Anthropic claims it’s just a documentation cleanup, actions like sending a cease-and-desist to OpenClaw and banning OpenCode from OAuth access suggest a shift towards requiring paid API keys for commercial development. This could signal the end of affordable flat-rate AI subscriptions for building products. <a href="https://app.daily.dev/feed-by-ids?id=3lsZ5Vula&amp;id=FQxtQgg9R&amp;id=GYB8zPRsi" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>MySQL community challenges Oracle’s governance</strong>: The MySQL community, including engineers from Percona and MariaDB, is pushing Oracle to transfer MySQL governance to an independent non-profit foundation. Concerns include declining commit activity, a shrinking contributor base, lack of AI-era features, and an opaque roadmap. They point to PostgreSQL’s community-led success as a model, but analysts are skeptical if Oracle retains trademark control. <a href="https://app.daily.dev/posts/GY7vcPtOu" target="_blank" rel="noopener nofollow">Read more</a></li>
</ul>
<h2>Good to know</h2>
<ul>
<li><strong>AI coding tools: feelings vs facts</strong>: Wix Engineering’s 5-week experiment found that while developers felt more productive with AI coding tools like Cursor, actual measurable output gains were only significant with Claude Code, and only after developers learned to trust it. <a href="https://app.daily.dev/posts/qQhctEUPB" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>AI agent actions: what you see isn’t what you get</strong>: AI agents in Docker containers perform far more syscalls than their logs reveal, making kernel-level tracing with tools like Azazel essential for verifying their actions and detecting suspicious behavior. <a href="https://app.daily.dev/feed-by-ids?id=cbOVi24pM&amp;id=dtY8IcMQN&amp;id=iGzH52Vzz&amp;id=k9XThzl85&amp;id=qBKhvArRq" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Spotify uses multi-agent AI for ad planning</strong>: Spotify Engineering built a multi-agent AI system called Ads AI using Google’s Agent Development Kit and Vertex AI to generate optimized media plans in seconds, replacing manual work that took up to 30 minutes. <a href="https://app.daily.dev/posts/JXalp4JE2" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Redwood materials powers AI data centers</strong>: Redwood Materials, founded by former Tesla CTO JB Straubel, is rapidly expanding its energy storage business, repurposing EV batteries to meet the surging power demands of AI data centers, with Google and Nvidia backing its latest funding round. <a href="https://app.daily.dev/posts/1mD3jO5pQ" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>GitHub copilot now references code sources</strong>: GitHub’s Copilot now integrates code referencing, highlighting matching snippets from public repositories with links to original sources and licenses, though it doesn’t block suggestions matching public code. <a href="https://app.daily.dev/feed-by-ids?id=kcB1LfSD2&amp;id=lOJYvaow4&amp;id=rsFWrvtAL&amp;id=vDLyK99UX" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Google spanner gets cassandra query language support</strong>: Google Cloud now offers general availability for native Cassandra Query Language (CQL) API support on Spanner, allowing existing Apache Cassandra applications to migrate with minimal code changes and leverage Spanner’s global ACID transactions and strong consistency. <a href="https://app.daily.dev/posts/CKaLROkeW" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>BigQuery automates embedding generation</strong>: BigQuery now supports autonomous embedding generation, automatically maintaining embedding columns on tables based on source columns, simplifying semantic search without manual pipeline management. <a href="https://app.daily.dev/posts/BhFMZ3Oac" target="_blank" rel="noopener nofollow">Read more</a></li>
</ul>
<h2>Bullshit police</h2>
<ul>
<li><strong>The ‘AI slop’ is real, and it’s coming for your open source</strong>: Apparently, AI-generated code is flooding open-source projects with ‘low-quality pull requests,’ creating an ‘AI slop’ that overwhelms maintainers. InfoWorld is very concerned about this ‘verification collapse’ and the ‘dangerous asymmetry’ where code generation is free but review costs aren’t. Who knew giving everyone a magic code-writing wand would lead to more… code? <a href="https://app.daily.dev/posts/Ixa8rNXuv" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Agentic AI: the new career risk, apparently</strong>: A new report suggests that AI tools like Kiro and Claude Code are making non-adoption a ‘career risk’ for software engineers. Because, you know, if you’re not using AI to write your code, you’re basically obsolete. Never mind that ‘vibe coding’ without solid engineering principles leads to ‘spaghetti code’ and bugs. Just keep prompting, I guess, and hope the AI doesn’t publish a hit piece on you. <a href="https://app.daily.dev/feed-by-ids?id=037a5YIjx&amp;id=4CjaWv3pP&amp;id=7XxAvY7Df&amp;id=Bl6akHr3q&amp;id=DqRWRtC6t&amp;id=F8eGl7g0G&amp;id=LCg5ueVO7&amp;id=SBASpAo7A&amp;id=UJqBwwcg1&amp;id=Uw1m3wdti" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>AI’s latest innovation: narco submarines</strong>: The MIT Technology Review is reporting on autonomous narco submarines using off-the-shelf tech like Starlink and autopilots for drug smuggling. Because, of course, the first thing we do with advanced AI and robotics is… make drug running more efficient. Meanwhile, Google DeepMind is ‘rigorously evaluating the moral behavior of LLMs.’ Priorities, people, priorities. <a href="https://app.daily.dev/feed-by-ids?id=EYFxMvhsd&amp;id=EwwCVZLpO" target="_blank" rel="noopener nofollow">Read more</a></li>
</ul>
