<h2>Must know</h2>
<ul>
<li><strong>Google releases powerful new gemini 3.1 pro model</strong>: Google has launched Gemini 3.1 Pro in public preview, touting it as their most powerful agentic and coding model yet. It boasts a massive 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 is available across various platforms, including GitHub Copilot, Google AI Studio, and Vertex AI, with a new configurable thinking level parameter for complex problem-solving. <a href="https://app.daily.dev/posts/PjE6Xo1XG" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Anthropic’s policy shift impacts third-party tools</strong>: Anthropic updated its Claude Code documentation, restricting the use of OAuth tokens from Pro/Max subscription accounts in third-party tools, causing significant backlash from the developer community. 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 move towards requiring paid API key authentication for commercial development. This signals a potential end to affordable flat-rate AI subscriptions for developers building businesses. <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 demands independent governance from oracle</strong>: Over 248 database professionals, including engineers from major players like Percona and MariaDB, have signed an open letter urging Oracle to transfer MySQL’s 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, highlighting growing fragmentation and the need for a more open development process. <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 workflows</strong>: Agentic AI is shifting software development from prompt-response to autonomous agents that plan, execute, and adapt, with tools like Kiro and Claude Code increasing engineer productivity, though requiring developers to evolve into ‘Agent Orchestration Engineers’. <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>New 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 introduced Epic, an AI workforce management platform providing shared long-term memory for AI coding agents. <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 and energy demands surge</strong>: Cisco and NVIDIA are partnering on 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 to meet the escalating energy demands of AI data centers. <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>Redwood materials expands energy storage for AI data centers</strong>: Redwood Materials, backed by Google and Nvidia, is rapidly expanding its energy storage business unit, repurposing EV batteries to meet the surging power demands of AI data centers, which face significant grid connection delays. <a href="https://app.daily.dev/posts/1mD3jO5pQ" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>GitHub copilot integrates code referencing</strong>: GitHub’s Copilot now integrates code referencing, highlighting matching code snippets from public GitHub repositories in session logs and providing 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>Amazon key reduces onboarding with event-driven platform</strong>: Amazon Key’s engineering team re-architected their event platform to a centralized, event-driven system built on Amazon EventBridge, reducing service onboarding from 48 hours to 4 and integration time from 40 hours to 8. <a href="https://app.daily.dev/posts/VWAzA59hA" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Workload identity federation now 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 instead of static credentials, with full CRUD support via API and Terraform integration. <a href="https://app.daily.dev/posts/ekx9IIs1v" target="_blank" rel="noopener nofollow">Read more</a></li>
</ul>
<h2>Bullshit police</h2>
<ul>
<li><strong>AI productivity experiment reveals ‘AI curse’</strong>: Wix Engineering’s 5-week experiment found that while Claude Code significantly boosted individual developer output, tools like Cursor showed almost no measurable gain despite users ‘feeling’ more productive. Apparently, this individual productivity explosion is creating an ‘AI Curse’ where organizational coordination, CI/CD, and review processes become the new bottlenecks. So, we’re just shifting the mess around, not actually cleaning it up. <a href="https://app.daily.dev/posts/qQhctEUPB" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>AI code quality and production visibility concerns</strong>: Articles are now warning that AI coding tools are shipping JavaScript faster than developers can review it, leading to ‘subtle production bugs’ and ‘AI slop’ in open-source repos. The proposed solution? ‘Production visibility’ tools to capture real-browser errors. So, instead of AI making code better, it’s making it worse, and now we need more tools to find the new kinds of bugs it introduces. Progress! <a href="https://app.daily.dev/feed-by-ids?id=2Xamq3Lpe&amp;id=IR0A3b0XK" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>AI agent demos fail in production due to telemetry overload</strong>: Apparently, all those slick AI agent demos often fail in production because ‘unbounded telemetry data’ overwhelms context windows. Experts are now saying vector memory alone isn’t enough, and we need ‘schema-driven knowledge graphs’ for ‘causal reasoning’ and ‘observability chaos.’ So, the problem isn’t the AI, it’s just that we haven’t built an entire new data infrastructure layer to support its basic functionality. Got it. <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>
</ul>
