<h2>Must know</h2>
<ul>
<li><strong>Google releases gemini 3.1 pro, claims top spot in agentic AI</strong>: Google has launched Gemini 3.1 Pro in public preview, touting it as their most powerful agentic and coding model. It scored 77.1% on the ARC-AGI-2 benchmark, significantly outperforming previous versions and competitors like Anthropic’s Opus 4.6 and OpenAI’s GPT-5.2. The model features a 1 million token context window, multimodal inputs, and a new configurable thinking level, with a ‘high’ mode capable of solving complex problems like math olympiads. Early adopters report gains in problem-solving and code generation, positioning it as a major contender in the AI agent space. <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 OAuth token usage from Pro/Max subscription accounts in third-party tools, causing backlash from developers. While Anthropic claims it’s a documentation cleanup, official legal documents state these tokens are for Claude Code and <a href="http://Claude.ai" target="_blank" rel="noopener nofollow">Claude.ai</a> only, requiring API key authentication for developers building products. This move, coupled with cease-and-desist actions against tools like OpenClaw, suggests a shift away from affordable flat-rate AI subscriptions for commercial development, pushing developers towards paid API keys via the Agent SDK. <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 proposes a foundation model with a technical steering committee, citing PostgreSQL’s community-led success as a contrast. This highlights a significant challenge to the foundational database’s future in an evolving tech landscape. <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>: AI tools like Kiro and Claude Code are shifting software development from prompt-response to autonomous agents that plan, execute, and adapt, with McKinsey data indicating 23% of organizations are scaling agentic AI. <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>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 launched the Secure AI Factory, a modular architecture for AI compute, networking, storage, and security, while SoftBank plans a $33 billion natural gas power plant to meet the massive 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>AI security tools and vulnerability management advance</strong>: New AI-powered SAST tools combine deterministic rules and LLM-based reasoning to improve vulnerability detection, and BARBIE, a new backdoor detection method for deep learning models, was presented at NDSS 2025. <a href="https://app.daily.dev/feed-by-ids?id=2TCivC1ph&amp;id=9x8XnEM3d&amp;id=CBTU3pEb0&amp;id=zsAWpz3GM" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>AI agent security, memory, and performance challenges</strong>: Azazel, an open-source eBPF-based runtime tracer, captures kernel-level events for AI agents, while schema-driven knowledge graphs are proposed for persistent, self-improving memory to handle observability chaos in production AI agents. <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>
<li><strong>AI’s impact on productivity and employment</strong>: A European study found AI adoption increases labor productivity by 4% with no short-run job losses, while an Indian survey suggests AI complements high-skill workers, challenging doomsday narratives about job displacement. <a href="https://app.daily.dev/feed-by-ids?id=iZlTkI8Me&amp;id=tiBubXcsb&amp;id=yOsJJlumD" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>AI code quality and production visibility concerns</strong>: AI coding tools accelerate JavaScript shipping, outpacing developer review capacity and introducing subtle production bugs, leading to calls for enhanced production visibility to understand the user impact of these new bug types. <a href="https://app.daily.dev/feed-by-ids?id=2Xamq3Lpe&amp;id=IR0A3b0XK" target="_blank" rel="noopener nofollow">Read more</a></li>
</ul>
<h2>Bullshit police</h2>
<ul>
<li><strong>Claude pro integrates with microsoft powerpoint</strong>: Oh joy, Claude Pro now integrates with Microsoft PowerPoint. Because what we really needed was another AI to help us make even more bullet-point-laden presentations. Apparently, it uses ‘connectors’ to pull context from ‘daily tools,’ which sounds suspiciously like it just reads your stuff and tries to make slides. The promotional material claims Claude Pro is ‘getting bigger every week,’ which is a classic sign of a feature looking for a problem. <a href="https://app.daily.dev/posts/oLQ4V0bPw" 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, has graced us with an ‘AI Management Dashboard’ to control AI crawler traffic. Because apparently, AI crawlers are so special they need their own analytics category, distinct from all the other bots. It promises ‘forensic drilldowns’ and ‘clear audit trails,’ which sounds like a lot of fancy words for a glorified log viewer. Just another tool to manage the AI chaos we’re all creating. <a href="https://app.daily.dev/posts/zxo3wKZNk" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Trajectory explorer tool announcement</strong>: Some guy named <code>benhylak</code> announced <code>Trajectory Explorer</code> on Twitter, claiming it’s ‘the first sane way to navigate agent traces.’ Because before this, we were all just blindly stumbling through our AI agents’ decisions, apparently. It’s a product launch, so naturally, details are scarce, but the hype is already in full swing. Can’t wait for the next ‘sane’ way to do something that probably wasn’t that insane to begin with. <a href="https://app.daily.dev/posts/AzPQg0qJn" target="_blank" rel="noopener nofollow">Read more</a></li>
</ul>
