<h2>Must know</h2>
<ul>
<li><strong>Salesforce phases out Heroku, pivots to AI</strong>: Salesforce is moving its Platform-as-a-Service (PaaS) Heroku into a ‘sustaining engineering’ phase, effectively halting new enterprise contracts and signaling a potential phase-out. This strategic shift aligns with Salesforce’s broader pivot towards AI-led growth, leaving existing Heroku users to consider migration to alternatives like Render or Vercel. The move highlights the increasing pressure on cloud platforms to align with AI strategies, as reported by various analysts. <a href="https://app.daily.dev/posts/13PGZg8yI" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>AI agents develop C compiler for Linux kernel</strong>: Anthropic researchers used 16 Claude Opus 4.6 AI agents to autonomously build a Rust-based C compiler capable of compiling the Linux kernel. While the experiment consumed $20,000 in API costs and produced 100,000 lines of code, the resulting compiler is not production-ready due to extreme inefficiency and subpar Rust quality. This demonstrates AI’s capability for complex software development, but also its current limitations in generating optimized, production-grade code, as detailed in the research. <a href="https://app.daily.dev/feed-by-ids?id=qhPl6UlEO&amp;id=9jEdtbewr" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>EU accuses Meta of WhatsApp AI antitrust violation</strong>: The European Commission has formally accused Meta of violating EU competition rules by blocking third-party AI assistants like ChatGPT and Microsoft Copilot from WhatsApp. Following policy changes in October 2025, Meta AI became the sole AI assistant allowed on the platform, leading the EU regulator to consider emergency measures to restore competitor access. Meta disputes the claims, arguing that AI assistants have multiple distribution channels. <a href="https://app.daily.dev/posts/w9RLQf8YH" target="_blank" rel="noopener nofollow">Read more</a></li>
</ul>
<h2>Good to know</h2>
<ul>
<li><strong>AI’s dual impact on software development</strong>: AI is fundamentally changing software development, leading to declining engagement in open-source projects due to ‘vibe coding’ and introducing new security vulnerabilities through hallucinated packages, while also shifting developer roles towards higher-level supervision. <a href="https://app.daily.dev/feed-by-ids?id=0CiZDAswQ&amp;id=0hMgfxxrs&amp;id=7G6kpg12b&amp;id=8yRieynOP&amp;id=KCC82Mw6D&amp;id=d4YFQW1JT&amp;id=eD5zl44R8&amp;id=gVU0l8bpQ" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>AI agents influence programming language design</strong>: The rise of AI agents is prompting a re-evaluation of programming language design, with new languages anticipated to prioritize explicit typing, avoid whitespace-based syntax, and support deterministic testing for optimal agent workflows. <a href="https://app.daily.dev/posts/PKgkpakDW" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Python ecosystem updates and security funding</strong>: Python 3.15 alpha shows JIT compiler gains, pandas 3.0 introduces breaking changes, and the Python Software Foundation received $1.5 million from Anthropic for security infrastructure improvements. <a href="https://app.daily.dev/feed-by-ids?id=0nZHWqg8R&amp;id=K4uvLmggX&amp;id=PuaaXOtow" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Valkey performance enhancements post-Redis fork</strong>: Valkey, a Redis fork, is undergoing significant performance improvements in versions 8 and 9, including a redesigned hash table architecture and memory optimizations, resulting in up to 40% memory reduction for small key-value workloads while maintaining Redis 7.2 compatibility. <a href="https://app.daily.dev/posts/BJPWCn9Zl" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>GitHub apps get enterprise teams API access</strong>: GitHub Apps can now access Enterprise Teams API endpoints with fine-grained permissions, allowing programmatic management of teams at scale without personal access tokens, enhancing security and automation. <a href="https://app.daily.dev/posts/5IGPhAfLF" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Real-time collaboration with Django-channels and Yjs</strong>: A real-time collaborative editing solution was implemented using Django-Channels with Yjs CRDTs, leveraging y-websocket and pycrdt-websocket for server-side operations without NodeJS dependencies, and custom JWT authentication. <a href="https://app.daily.dev/posts/5wWt4VRq7" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>NVIDIA automates LLM inference optimization</strong>: NVIDIA TensorRT LLM AutoDeploy, a new beta feature, automates the compilation of PyTorch models into optimized inference engines for LLMs, significantly reducing deployment time from weeks to days. <a href="https://app.daily.dev/posts/QkLXaUNML" target="_blank" rel="noopener nofollow">Read more</a></li>
</ul>
<h2>Bullshit police</h2>
<ul>
<li><strong>Anthropic’s Claude co-work plugin ‘wipes out’ $400 billion in tech stocks</strong>: Anthropic’s new Claude Co-work plugins, designed for industry-specific tasks, supposedly ‘wiped out $400 billion in tech stocks’ by ‘triggering market concerns’ about the SaaS business model. Because, clearly, a new AI feature is solely responsible for massive market shifts, and not, you know, the usual volatility or other economic factors. The article’s title is a masterclass in definitive, uncorroborated market analysis. <a href="https://app.daily.dev/posts/QnkkrfiEd" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>AI therapy chatbots: the future of mental health?</strong>: AI therapy chatbots like Wysa and Woebot are apparently ‘rising to address mental health,’ while a University of Oxford study found AI chatbots don’t improve medical advice compared to search engines. Users struggled to provide relevant info, and models gave inconsistent recommendations. So, we’re trading inconsistent medical advice for inconsistent mental health advice? Sounds like a solid plan for societal well-being. <a href="https://app.daily.dev/feed-by-ids?id=jmvy6d70v&amp;id=nTkvD3DKT" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>AI-generated code: 56% faster, 19% slower, 96% untrusted</strong>: AI coding assistants are a real head-scratcher: one study says they make developers 56% faster, another claims 19% slower due to review overhead. Meanwhile, 96% of engineers don’t trust AI-generated code, yet half of them commit it without verification. It’s almost as if the ‘productivity’ gains are just a fancy way of saying ‘more work for someone else to fix later,’ all while we pretend to be efficient. <a href="https://app.daily.dev/feed-by-ids?id=OTko9hG2g&amp;id=QxTS3kJG9&amp;id=t4jX0rY0U&amp;id=tbBooOAn7&amp;id=vO7uBsZ50" target="_blank" rel="noopener nofollow">Read more</a></li>
</ul>
