Atlassian

Changelog

Cisco

Google Cloud

Hacker News

InfoWorld

<h2>Must know</h2>
<ul>
<li><strong>Google releases Gemini 3.1 Pro with major performance gains</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 the ARC-AGI-2 benchmark, a significant leap from its predecessor’s 31.1%, and leads most coding benchmarks. This model supports multimodal inputs and introduces a configurable thinking level, intensifying the AI model race with competitors like Anthropic and OpenAI also pushing new releases. <a href="https://app.daily.dev/posts/PjE6Xo1XG" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>AI agent adoption reshapes software development</strong>: AI coding tools are fundamentally changing developer roles, shifting focus from writing code to reviewing and orchestrating AI agents, with 93% of developers now using AI assistants. While productivity gains are reported, concerns are rising about ‘vibe-coding’ in critical systems and subtle production bugs introduced by AI-generated code, highlighting the need for robust testing and observability. New roles like Agent Orchestration Engineer are emerging, emphasizing the importance of engineering maturity for reliable AI application deployment. <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 by community</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. Citing declining commit activity, a shrinking contributor pool, and a lack of AI-era features, the community seeks to revitalize MySQL development and address growing fragmentation across its forks. This move highlights ongoing concerns about Oracle’s control over the popular database. <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>Red Hat Lightspeed MCP for RHEL image creation</strong>: Red Hat Lightspeed Model Context Protocol (MCP) is enabling natural language-driven RHEL image creation and management, allowing users to create blueprints, apply security profiles, and push images to cloud providers through conversational prompts. <a href="https://app.daily.dev/posts/CtJea7QHd" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Spotify uses multi-agent AI for smarter advertising</strong>: Spotify Engineering built an Ads AI system with specialized agents to optimize media planning workflows, generating optimized media plans in seconds using Google’s Agent Development Kit and Vertex AI. <a href="https://app.daily.dev/posts/JXalp4JE2" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>BigQuery now supports autonomous embedding generation</strong>: Google Cloud’s BigQuery now automatically maintains embedding columns on tables and simplifies 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>MotherDuck launches ‘Dives’ for AI-driven data visualization</strong>: MotherDuck’s new ‘Dives’ feature allows AI agents like Claude and Gemini to build live, shareable data visualizations with real-time SQL queries, 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>Mirai raises seed round for on-device AI inference engine</strong>: Mirai, a London-based startup, secured $10M to develop an inference engine in Rust for Apple Silicon, claiming up to 37% faster model generation speed for on-device assistants, transcription, and translation. <a href="https://app.daily.dev/posts/EN0GXQWa8" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Link11 launches AI management dashboard for crawler traffic</strong>: Link11, an IT security provider, released an AI Management Dashboard offering dedicated visibility and control over AI crawler traffic, enabling companies to identify specific AI systems accessing their content and enforce access policies. <a href="https://app.daily.dev/posts/zxo3wKZNk" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Vitest 3.x significantly outperforms Jest in testing benchmarks</strong>: Vitest 3.x shows 5.6x faster cold starts and 28x faster watch mode re-runs compared to Jest 30 in a 50,000-test monorepo, attributed to its Vite-based transform pipeline and shared module caching. <a href="https://app.daily.dev/posts/60rKeST9w" target="_blank" rel="noopener nofollow">Read more</a></li>
</ul>
<h2>Bullshit police</h2>
<ul>
<li><strong>AI agent governance: a wild west of closed-source models</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, we’re just supposed to trust these black boxes, many of which are making critical decisions, without any real transparency or accountability? What could possibly go wrong? <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>
<li><strong>AI’s text-trap: commoditizing your SaaS, one generic text block at a time</strong>: Apparently, text-only AI assistants are going to ‘commoditize SaaS products’ by stripping away carefully crafted user experiences. The solution? More ‘rich output,’ ‘UI as input,’ and ‘co-creation’ to avoid the ‘generic mediocrity of generative UI.’ Because, you know, the problem isn’t the AI, it’s that we’re not making it generate <em>prettier</em> text boxes. Groundbreaking. <a href="https://app.daily.dev/posts/Nz3KEZcnf" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>AI agent for navigating traces: the ‘first sane way’ to inspect 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 before this, we were all just wildly guessing what our AI was doing, right? It’s always the ‘first sane way’ until the next ‘first sane way’ comes along. <a href="https://app.daily.dev/posts/AzPQg0qJn" target="_blank" rel="noopener nofollow">Read more</a></li>
</ul>
