<h2>Must know</h2>
<ul>
<li><strong>Google rolls out gemini 3.1 pro</strong>: Google has launched Gemini 3.1 Pro in public preview, touting it as their most powerful agentic and coding model. It boasts a 1 million token context window and scored 77.1% on the ARC-AGI-2 benchmark, significantly outperforming its predecessor and competitors like Anthropic’s Opus 4.6 and OpenAI’s GPT-5.2 on this specific metric. The model supports multimodal inputs and introduces a configurable thinking level parameter, with a ‘high’ mode capable of solving complex problems. <a href="https://app.daily.dev/posts/PjE6Xo1XG" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Anthropic claude code policy shift causes developer backlash</strong>: Anthropic updated its Claude Code documentation, restricting OAuth token usage from Pro/Max subscription accounts in third-party tools, leading to community uproar. While Anthropic claims it’s a documentation cleanup, developers relying on these tokens for tools like OpenClaw and NanoClaw are concerned. Official legal documents state API key authentication via Claude Console is required for developers building products, suggesting an end to affordable flat-rate AI subscriptions for commercial development. <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 governance under oracle challenged 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. Concerns include declining commit activity, a shrinking contributor pool, lack of AI-era features, and an opaque roadmap. The community points to PostgreSQL’s success with community-led governance as a model, though analysts are skeptical about resolving power dynamics 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 agents reshape software development</strong>: AI tools like Kiro and Claude Code are shifting the productivity baseline for software engineers, with Visual Studio shipping built-in and custom agent frameworks, and developers’ roles evolving into Agent Orchestration Engineer and LLMOps Specialist. <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>Wix engineering experiment shows mixed AI productivity results</strong>: A 5-week experiment at Wix Engineering found Claude Code significantly boosted output after developers built trust, but Cursor showed almost no measurable productivity lift despite users feeling more productive, highlighting an ‘AI Curse’ where individual gains create organizational bottlenecks. <a href="https://app.daily.dev/posts/qQhctEUPB" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>AI memory and agent management solutions emerge</strong>: Mengram launched an open-source AI memory API with semantic, episodic, and procedural memory types, while Reload’s Epic platform provides shared long-term memory for AI coding agents, addressing current AI agent memory and collaboration limitations. <a href="https://app.daily.dev/feed-by-ids?id=0tYkAozTe&amp;id=MQxvV1b9u" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>AI infrastructure development and energy demands</strong>: Cisco and NVIDIA launched the Secure AI Factory, a modular architecture for AI compute, networking, storage, and security, while SoftBank plans a massive 9.2 gigawatt natural gas power plant, highlighting the significant infrastructure investments required for AI. <a href="https://app.daily.dev/feed-by-ids?id=IYsqAR4Cy&amp;id=VBQEBjah6" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>AI and robotics industrial adoption accelerates</strong>: Toyota deployed seven Agility Robotics Digit humanoid robots at its RAV4 plant, and NVIDIA’s survey reports 90% of telecom respondents see AI increasing revenue and reducing costs, indicating AI and robotics are moving from pilot to industrial deployment. <a href="https://app.daily.dev/feed-by-ids?id=WnnenuOfb&amp;id=p1OVwpWSR" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>OpenAI expands strategically in India</strong>: OpenAI partnered with Pine Labs to embed AI APIs into fintech infrastructure and with Reliance to integrate AI-powered conversational search into JioHotstar, alongside plans for data center partnerships with Tata Group and new offices in Mumbai and Bengaluru. <a href="https://app.daily.dev/feed-by-ids?id=JlixRkBbI&amp;id=JodTqUZtB" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Node.js ecosystem updates and performance benchmarks</strong>: Node Weekly Issue 612 covers <code>node-caged</code> for halving Node.js memory usage, npm v11.10.0, and BrowserPod 1.0, while a benchmark shows Vitest 3.x significantly outperforms Jest 30 in cold starts and watch mode re-runs. <a href="https://app.daily.dev/feed-by-ids?id=Mm7b9KCfF&amp;id=T3G4VUzdk" target="_blank" rel="noopener nofollow">Read more</a></li>
</ul>
<h2>Bullshit police</h2>
<ul>
<li><strong>AI slop is flooding open source, creating a ‘verification collapse’</strong>: Apparently, AI-generated code is so good it’s now ‘flooding open source projects with low-quality pull requests,’ according to InfoWorld. Maintainers are ‘overwhelmed by AI slop’ that looks professional but contains ‘subtle logic errors, security flaws, or unmaintainable complexity.’ So, the free code generation is great, but the free review burden? Not so much. Who knew free lunch came with a side of technical debt? <a href="https://app.daily.dev/posts/Ixa8rNXuv" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>AI agents in docker containers: more syscalls than you can shake a stick at</strong>: A new tool, Azazel, reveals AI agents in Docker containers execute ‘many more syscalls’ than their application logs let on. Apparently, a ‘simple task’ involves 47 process_exec, 312 file_open, and 8 net_connect. Because, you know, AI agents are just too busy being agentic to bother logging their actual kernel-level shenanigans. Good thing we have eBPF to tell us what they’re <em>really</em> doing. <a href="https://app.daily.dev/posts/cbOVi24pM" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Don’t use next.js if you’re building for the long term</strong>: Someone out there is bravely stating the obvious: ‘Don’t Use Next.js If You’re Building for the Long Term.’ The argument? ‘Upgrade churn, tight coupling to Vercel’s ecosystem, and implicit behavior traps create compounding risks.’ Shocking, I know. It’s almost like chasing the latest shiny framework might not be the best strategy for stability. Who would’ve thought? <a href="https://app.daily.dev/posts/0vMYzSZd0" target="_blank" rel="noopener nofollow">Read more</a></li>
</ul>
