<h2>Must know</h2>
<ul>
<li><strong>Google releases gemini 3.1 pro preview</strong>: Google just dropped Gemini 3.1 Pro in public preview, touting it as their latest agentic coding and state-of-the-art reasoning model. It scored 77.1% on ARC-AGI-2, a huge jump from 31.1% for Gemini 3 Pro, and leads most coding benchmarks. This model supports multimodal inputs and offers a 1 million token context window, positioning it as a strong contender in the AI model race, especially with its competitive pricing against Anthropic’s Opus 4.6. <a href="https://app.daily.dev/posts/PjE6Xo1XG" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>AI agents are changing software development</strong>: AI coding tools are shifting developer roles from writing code to reviewing and orchestrating AI agents, with 93% of developers now using AI assistants. While productivity gains are reported, like 60-70% debug time reduction, AI-generated code can introduce subtle bugs and security risks, as seen in a $1.78 million DeFi protocol loss due to a Claude Opus 4.6 co-authored misconfiguration. This highlights the need for new developer roles focused on agent orchestration and robust engineering maturity. <a href="https://app.daily.dev/feed-by-ids?id=037a5YIjx&amp;id=0Q86tGAco&amp;id=0tYkAozTe&amp;id=1v9gMZMqJ&amp;id=2LocixQWJ&amp;id=2WWQ65Wm6&amp;id=2Xamq3Lpe&amp;id=3Kdb5qCcA&amp;id=4CjaWv3pP&amp;id=6KUpHm9AN" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>MySQL governance under scrutiny</strong>: Over 248 database professionals, including engineers from Percona and MariaDB, have signed an open letter urging Oracle to transfer MySQL governance to an independent non-profit foundation. They cite declining commit activity, a shrinking contributor pool, lack of AI-era features, and an opaque roadmap as key concerns. The community hopes to revitalize MySQL development and ensure its future relevance, drawing parallels to PostgreSQL’s successful community-led model. <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>MotherDuck launches dives for AI agents</strong>: MotherDuck launched ‘Dives’ in public preview, allowing AI agents like Claude and Gemini to build live, shareable data visualizations with real-time SQL queries against MotherDuck, aiming to shift focus from query performance to answering business questions. <a href="https://app.daily.dev/posts/5vuQOT9lb" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Google Cloud adds native CQL API support to spanner</strong>: Google Cloud now offers general availability for native CQL (Cassandra Query Language) API support on Spanner, enabling Cassandra applications to migrate with minimal code changes and leverage Spanner’s global ACID transactions and 99.999% availability. <a href="https://app.daily.dev/posts/CKaLROkeW" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Red Hat lightspeed uses model context protocol for RHEL image creation</strong>: Red Hat Lightspeed is leveraging Model Context Protocol (MCP) to enable natural language-driven RHEL image creation and management, allowing users to create blueprints, apply security profiles, and push images to cloud providers via conversational prompts. <a href="https://app.daily.dev/posts/CtJea7QHd" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Tailscale workload identity federation is generally available</strong>: Tailscale’s workload identity federation is now generally available, allowing CI/CD pipelines and cloud workloads to authenticate using short-lived OIDC tokens, eliminating static credentials and integrating with Terraform and Kubernetes. <a href="https://app.daily.dev/posts/ekx9IIs1v" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>BigQuery now supports autonomous embedding generation</strong>: BigQuery has introduced autonomous embedding generation, automatically maintaining embedding columns on tables and simplifying semantic search with a new <code>AI.SEARCH</code> function, integrating with Vertex AI embedding models. <a href="https://app.daily.dev/posts/BhFMZ3Oac" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Node.js memory usage halved with node-caged</strong>: Cloudflare, Igalia, and the Node.js project collaborated on <code>node-caged</code>, a Docker image with V8 pointer compression that can halve Node.js memory usage in production, as detailed in Node Weekly Issue 612. <a href="https://app.daily.dev/posts/Mm7b9KCfF" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Rust participates in google summer of code 2026</strong>: The Rust Project is participating in Google Summer of Code (GSoC) for the third consecutive year, inviting contributors to work on project ideas with mentors available on the Rust Zulip #gsoc stream. <a href="https://app.daily.dev/posts/FdD6cw6Ce" target="_blank" rel="noopener nofollow">Read more</a></li>
</ul>
<h2>Bullshit police</h2>
<ul>
<li><strong>Claude pro integrates with microsoft powerpoint, because reasons</strong>: Anthropic announced Claude Pro now integrates with Microsoft PowerPoint, allowing users to pull ‘context’ from daily tools directly into presentations. Because what every top-tier executive needs is an AI to help them make more slides, rather than, you know, actual insights. The promotional framing suggests Claude Pro is ‘getting bigger every week,’ which is great for PowerPoint, I guess. <a href="https://app.daily.dev/posts/oLQ4V0bPw" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Trajectory explorer: The ‘first sane way’ to inspect AI agent decisions</strong>: @benhylak announced Trajectory Explorer, a new tool for navigating AI agent traces, claiming it’s the ‘first sane way’ to inspect AI agent decisions. Because apparently, before this, we were all just wildly guessing what our AI was doing. The promotional language is strong, but let’s see if it actually delivers on making AI agent debugging ‘sane’ or just adds another layer of complexity. <a href="https://app.daily.dev/posts/AzPQg0qJn" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>AI agent governance: More closed-source, less transparency</strong>: MIT CSAIL’s 2025 AI Agent Index found 23 of 30 AI agents are closed-source, with only 4 of 13 frontier-autonomy agents disclosing safety evaluations. So, while everyone’s talking about AI governance, the industry seems to be doubling down on opacity. It’s almost as if ‘accountability’ is just a buzzword until it’s time to open up the black box. <a href="https://app.daily.dev/feed-by-ids?id=0j71k63Zt&amp;id=3vnyKPx4I&amp;id=4YRSUVlsV&amp;id=8NUiIgn7m&amp;id=DqRWRtC6t&amp;id=EYFxMvhsd&amp;id=EwwCVZLpO&amp;id=HjxE0TjWy&amp;id=J8WeyapTq&amp;id=Nz3KEZcnf" target="_blank" rel="noopener nofollow">Read more</a></li>
</ul>
