<h2>Must know</h2>
<ul>
<li><strong>AI coding tools: the good, the bad, and the ugly</strong>: Salesforce deployed Cursor AI to 20,000 engineers, reporting double-digit improvements in cycle time and code quality, with 90% adoption. However, a hands-on review of OpenCode, a local LLM code generation tool, found it underperforms on complex tasks and produces verbose, generic code, requiring expert oversight. Meanwhile, Tailwind Labs laid off 75% of its developers, attributing it to AI-assisted coding reducing web traffic to documentation sites, impacting open-source project monetization. This highlights the mixed impact of AI on developer productivity and the sustainability of open-source projects, as reported by Salesforce and anonymous sources. <a href="https://app.daily.dev/feed-by-ids?id=7KqSlBkW9&amp;id=Bph7neqjl&amp;id=EHuoSn58k&amp;id=PBAzOX1yL" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Cybersecurity: ransomware, phishing, and supply chain attacks on the rise</strong>: Ingram Micro disclosed a ransomware attack by Safepay affecting over 42,000 employees, while the Everest group claims to have leaked 72.7 million Under Armour customer accounts. LastPass users are targeted by a phishing campaign using fake maintenance emails to steal master passwords. Additionally, North Korean threat group PurpleBravo used fake job interviews to deliver malware via Visual Studio Code projects, targeting 3,136 IP addresses. These incidents underscore the escalating and diverse cyber threats, as reported by multiple security outlets. <a href="https://app.daily.dev/feed-by-ids?id=rDW8Xu1hd&amp;id=0UoXRJ0Z4&amp;id=TCuZthJYG&amp;id=btyfpzMTN&amp;id=sGrgwY6TO" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>AI’s impact on civic institutions and societal structures</strong>: A Boston University paper argues that AI systems are eroding the foundations of democratic societies by sidelining human judgment and weakening institutional checks and balances. This perspective contrasts with Microsoft CEO Satya Nadella’s view that AI sovereignty is about control over models trained with proprietary knowledge, not just data location. The debate highlights growing concerns about AI’s long-term societal effects and the need for careful consideration of its integration into critical systems, as discussed by Hackaday and the World Economic Forum. <a href="https://app.daily.dev/feed-by-ids?id=zxareKd2O&amp;id=n1gYMuFrF&amp;id=N5NjltfbY" target="_blank" rel="noopener nofollow">Read more</a></li>
</ul>
<h2>Good to know</h2>
<ul>
<li><strong>New AI wearables from Apple and OpenAI</strong>: Apple is developing an AI-powered wearable pin with dual cameras and three microphones for a 2027 launch, while OpenAI plans to release AI-powered earbuds, codenamed ‘Sweet Pea,’ in late 2026, featuring a custom 2-nanometer processor for local AI processing. <a href="https://app.daily.dev/feed-by-ids?id=fgglWe8Qn&amp;id=vOBU57mgd" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>AI-native CMS for live production websites</strong>: Strapi launched <a href="http://Fimo.ai" target="_blank" rel="noopener nofollow">Fimo.ai</a>, an AI-native content management system for live production websites, offering real-time visual editing, instant updates with rollbacks, and collaboration features, leveraging Neon’s Postgres database system for scalable architecture. <a href="https://app.daily.dev/posts/f5IhxeEO5" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Local LLM integration with Ollama and Python</strong>: Real Python released an interactive quiz on integrating local LLMs with Python using the Ollama library, focusing on setting up Ollama, pulling models, text generation, chat features, and tool calling for privacy-focused, offline workflows. <a href="https://app.daily.dev/posts/o2fGiPYCa" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>New open-source charting library with WebGPU</strong>: ChartGPU is an open-source TypeScript charting library that uses WebGPU for high-performance rendering of large datasets, supporting various chart types, interactions, and streaming data updates, with React bindings available. <a href="https://app.daily.dev/posts/bruP1nBn4" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Automated engineering standards enforcement platform</strong>: Earthly raised $5.7 million for Earthly Lunar, a guardrails engine that automates enforcement of engineering standards at the pull request level, offering over 100 pre-built guardrails and custom rule creation. <a href="https://app.daily.dev/posts/JYhaSPcPO" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>New open-source code snippet manager</strong>: Snipplle is an open-source code snippet manager built with Nuxt 4, TypeScript, and PostgreSQL, offering syntax highlighting, collections, workspaces, version control, CLI integration, and self-hosting options. <a href="https://app.daily.dev/feed-by-ids?id=AIlcmWOAv&amp;id=Lhaif3YU8" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>AI-powered voice-to-task feature for Todoist</strong>: Todoist introduced Ramble, an AI-enabled feature powered by Google’s Gemini 2.5 Flash Live model, which converts spoken language into structured tasks, extracting deadlines, priorities, and assignees in 38 languages. <a href="https://app.daily.dev/posts/83KOTqyoP" target="_blank" rel="noopener nofollow">Read more</a></li>
</ul>
<h2>Bullshit police</h2>
<ul>
<li><strong>AI-powered coding assistants: because we needed more ways to generate mediocre code</strong>: AionUi, a new open-source desktop app, promises a unified GUI for various CLI AI tools like Gemini CLI and Claude Code, positioning itself as a ‘Cowork platform.’ Because what developers really need is another layer of abstraction to churn out more code without actually understanding it. It’s almost as if the goal is to make everyone a ‘developer’ by simply clicking buttons, ensuring a future filled with even more generic, AI-generated solutions, as if the current crop wasn’t enough. <a href="https://app.daily.dev/posts/9FJW7wGug" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>AI for customer service: because humans were too efficient at being human</strong>: Bunq, Europe’s ‘second-largest neobank,’ deployed Finn, an AI customer support assistant using Amazon Bedrock and Anthropic’s Claude models, now managing 97% of support interactions with 82% fully automated. They claim to be Europe’s first AI-powered bank. Because, clearly, the human touch and careful verification in compliance-critical operations were just slowing things down. I’m sure nothing will go wrong when an LLM decides how to handle your banking inquiries. <a href="https://app.daily.dev/posts/mMFFFln2H" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>AI-driven supply chain security: because a fancy dashboard fixes everything</strong>: A Forrester study, conveniently commissioned by JFrog, claims unified software supply chain security platforms offer a 282% ROI and 80% faster vulnerability remediation. Meanwhile, Codenotary updated its <a href="http://SBOM.sh" target="_blank" rel="noopener nofollow">SBOM.sh</a> tool for AI applications, documenting datasets as ‘first-class supply chain artifacts.’ It’s almost as if throwing AI and a dashboard at the problem will magically make all our supply chain woes disappear, rather than addressing the fundamental lack of trust in software origins, as some argue. <a href="https://app.daily.dev/feed-by-ids?id=v3tvoqFP3&amp;id=xpDPqgvfa" target="_blank" rel="noopener nofollow">Read more</a></li>
</ul>
