<h2>Must know</h2>
<ul>
<li>Google rolls out Gemini 3.1 Pro preview: 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’s available across a bunch of platforms, from GitHub Copilot to Vertex AI, and boasts a 1 million token context window. This release intensifies the AI model race, with Google positioning it as a cheaper alternative to Anthropic’s Opus 4.6, despite some weaknesses in specific benchmarks. <a href="https://app.daily.dev/posts/PjE6Xo1XG" target="_blank" rel="noopener nofollow">Read more</a></li>
<li>MySQL governance under scrutiny: Over 248 database professionals, including engineers from Percona and MariaDB, are pushing Oracle to hand over MySQL governance to an independent non-profit. They’re concerned about declining commit activity, a shrinking contributor pool, lack of AI-era features, and an opaque roadmap. This move aims to revitalize MySQL development and ensure its future relevance, drawing parallels to PostgreSQL’s community-led success. <a href="https://app.daily.dev/posts/GY7vcPtOu" target="_blank" rel="noopener nofollow">Read more</a></li>
<li>AI agents are changing software development: 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, there are growing concerns about AI-generated code introducing subtle production bugs and the ‘AI slop problem’ in open source, where AI-generated PRs create extra review burden for maintainers. New roles like Agent Orchestration Engineer are emerging to manage this shift. <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>
</ul>
<h2>Good to know</h2>
<ul>
<li>Tailscale workload identity federation is GA: Tailscale’s workload identity federation is now generally available, letting CI/CD pipelines and cloud workloads authenticate with short-lived OIDC tokens instead of static credentials. <a href="https://app.daily.dev/posts/ekx9IIs1v" target="_blank" rel="noopener nofollow">Read more</a></li>
<li>Vitest 3.x crushes Jest 30 in performance: Vitest 3.x significantly outperforms Jest 30, showing 5.6x faster cold starts and 28x faster watch mode re-runs in a 50,000-test monorepo, while using 57% less memory. <a href="https://app.daily.dev/posts/T3G4VUzdk" target="_blank" rel="noopener nofollow">Read more</a></li>
<li>Apache Kafka 4.2.0 released with share groups: Apache Kafka 4.2.0 is out, bringing production-ready Share Groups (Kafka Queues), GA status for Kafka Streams server-side rebalance protocol, and improved observability and security features. <a href="https://app.daily.dev/posts/0DcTaIDm9" target="_blank" rel="noopener nofollow">Read more</a></li>
<li>Google Cloud Spanner now supports native CQL APIs: Google Cloud Spanner now offers general availability for native CQL (Cassandra Query Language) API support, enabling Cassandra application migration with minimal code changes and offering global ACID transactions. <a href="https://app.daily.dev/posts/CKaLROkeW" target="_blank" rel="noopener nofollow">Read more</a></li>
<li>Rust participates in Google Summer of Code 2026: The Rust Project is joining Google Summer of Code (GSoC) for the third consecutive year, with applications open from March 16 to March 31, 2026. <a href="https://app.daily.dev/posts/FdD6cw6Ce" target="_blank" rel="noopener nofollow">Read more</a></li>
<li>Moxy: a lightweight superset of C: Moxy, a lightweight superset of C, was introduced, transpiling to clean C11 and adding modern features like string/bool types, dynamic arrays, and Result types. <a href="https://app.daily.dev/posts/le91077M0" target="_blank" rel="noopener nofollow">Read more</a></li>
<li>KFC architecture blueprint for real-time data: The KFC Architecture Blueprint combines Kafka, Flink, and ClickHouse for end-to-end real-time data processing, suitable for fraud detection and clickstream analytics. <a href="https://app.daily.dev/posts/pAQO9MXVw" target="_blank" rel="noopener nofollow">Read more</a></li>
</ul>
<h2>Bullshit police</h2>
<ul>
<li>AI agent governance, bias, and economic impact concerns: MIT CSAIL’s 2025 AI Agent Index found most AI agents are closed-source and lack safety evaluations, while Microsoft used an AI-mangled, plagiarized Git workflow diagram. Apparently, the AI industry is too busy making money to bother with transparency or basic ethics. Who needs standards when you have ‘innovation’? <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>AI tools save time and money, says guy who uses 6 of them: One author claims six AI tools, including Perplexity and Claude, reduce his daily work to about two hours. Sure, because integrating six different AI tools, managing their outputs, and verifying their hallucinations is definitely less work than just doing the job yourself. It’s almost like these tools are designed to make you feel productive, not actually be productive. <a href="https://app.daily.dev/posts/X2M941duB" target="_blank" rel="noopener nofollow">Read more</a></li>
<li>AI agent CLI to control iOS and Android devices: Callstack released <code>agent-device</code>, a CLI tool for AI agents to control iOS and Android devices programmatically. Because what we really need is AI agents tapping around our phones, making decisions, and probably ordering things we don’t need. What could possibly go wrong with giving an AI agent full control over your mobile devices? <a href="https://app.daily.dev/posts/Ekfc0OFB7" target="_blank" rel="noopener nofollow">Read more</a></li>
</ul>