<h2>Must know</h2>
<ul>
<li><strong>NVIDIA and Lilly team up for AI drug discovery</strong>: NVIDIA and pharmaceutical giant Lilly are pouring $1 billion into an AI co-innovation lab to revolutionize drug discovery. Announced at the J.P. Morgan Healthcare Conference, this lab aims to merge computational modeling with experimental wet labs, leveraging NVIDIA’s AI infrastructure, including the DGX SuperPOD with B300 systems. The goal is to model biological complexities more accurately and accelerate the discovery of new molecules, as reported by NVIDIA. This initiative signals a major push for AI in biopharma, with NVIDIA also expanding its BioNeMo platform for RNA structure prediction and molecular design. <a href="https://app.daily.dev/posts/TcHQkCPbq" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>AI solves a long-standing math problem</strong>: GPT 5.2 Pro, an AI model, has reportedly solved Erdos problem #281, a mathematical challenge that has remained open until now. Mathematician Terence Tao, a highly credible source, stated this is “perhaps the most unambiguous instance” of AI solving an open mathematical problem. This marks a significant milestone for AI in mathematics, suggesting a new era for AI-assisted research, according to the provided information. <a href="https://app.daily.dev/posts/pRkd3CPEF" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Microsoft pushes unified AI governance and marketplace</strong>: Microsoft is positioning itself as a leader in unified AI governance, integrating responsible AI standards across its platforms like Foundry, Agent 365, Purview, Entra, and Defender. Recognized by IDC MarketScape, their platform offers centralized oversight for traditional ML, generative AI, and agentic AI, with features like real-time threat detection and compliance for over 100 regulatory frameworks. Additionally, the Microsoft Marketplace now boasts over 11,000 AI models and 4,000 AI applications, including pre-vetted solutions from Anthropic, OpenAI, Meta, and NVIDIA, aiming to accelerate enterprise AI adoption with a focus on security and compliance. <a href="https://app.daily.dev/feed-by-ids?id=VuTiPUrsF&amp;id=XdBf9IqBR" target="_blank" rel="noopener nofollow">Read more</a></li>
</ul>
<h2>Good to know</h2>
<ul>
<li><strong>NVIDIA boosts AI video and image generation on RTX</strong>: NVIDIA announced significant performance upgrades for AI video and image generation on RTX GPUs, including 3x faster 4K video generation and 60% VRAM savings via ComfyUI optimizations, and integration of Lightricks’ LTX-2 model for 4K video with audio. <a href="https://app.daily.dev/posts/JItr1KihQ" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>NVIDIA expands GeForce NOW to Linux and Fire TV</strong>: NVIDIA is expanding its GeForce NOW cloud gaming service with native apps for Linux PCs (Ubuntu 24.04+) and Amazon Fire TV devices, offering up to 5K resolution at 120fps or 1080p at 360fps streaming, powered by GeForce RTX 5080 GPUs. <a href="https://app.daily.dev/posts/5ceYZHV0v" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>AI agents automate particle accelerator operations</strong>: Lawrence Berkeley National Laboratory is using an LLM-driven AI Accelerator Assistant, running on NVIDIA H100 GPUs, to autonomously troubleshoot and manage their Advanced Light Source particle accelerator, reducing setup time by 100x. <a href="https://app.daily.dev/posts/ZzMUF9PNc" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>NVIDIA DRIVE powers Mercedes-Benz CLA and expands ecosystem</strong>: NVIDIA DRIVE AV software will be integrated into the new Mercedes-Benz CLA by late 2026, enabling advanced level 2 driver assistance, while NVIDIA also expanded its DRIVE Hyperion ecosystem at CES 2026 with partners like Bosch and Magna, targeting level 4 autonomous driving. <a href="https://app.daily.dev/feed-by-ids?id=kCsBTg7xZ&amp;id=pTEo8FtLn" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>ClickHouse acquires Langfuse for LLM tracing</strong>: ClickHouse has acquired Langfuse, an open-source platform for LLM tracing and evaluation, aiming to accelerate its development while maintaining open-source and self-hosting commitments, leveraging ClickHouse’s infrastructure for enhanced performance. <a href="https://app.daily.dev/feed-by-ids?id=O5KxwsHsv&amp;id=dliP781v1" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Black Basta ransomware leader targeted by EU and INTERPOL</strong>: Ukrainian and German authorities have identified Oleg Nefedov as the alleged leader of the Black Basta ransomware group, adding him to the EU Most Wanted and INTERPOL Red Notice lists, following a leak of internal chat logs that led to the group ceasing operations. <a href="https://app.daily.dev/posts/O0PAVFWl9" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>New JavaScript compiler Porffor targets embedded systems</strong>: Porffor is a new ahead-of-time JavaScript compiler that generates WebAssembly and native code, promising up to 50x smaller binaries and significant speed improvements for serverless, game engines, and embedded systems, though it currently lacks garbage collection and web APIs. <a href="https://app.daily.dev/feed-by-ids?id=xTAvQVuAU&amp;id=xU3ZYAZlY" target="_blank" rel="noopener nofollow">Read more</a></li>
</ul>
<h2>Bullshit police</h2>
<ul>
<li><strong>AI agents are just old automation with a new hat</strong>: Apparently, the ‘revolutionary’ concept of ‘AI agents’ is just a fancy rebrand of good old API and web automation, according to some technical folks. It seems the hype is more about investor FOMO and financial gain than any actual substantive new technology. So, next time someone gushes about ‘agentic automation,’ just remember it’s probably the same old script, just with more buzzwords and a bigger price tag. <a href="https://app.daily.dev/feed-by-ids?id=4rJ96gwiU&amp;id=J1BKrq2u6&amp;id=JE89HgTw1&amp;id=JItr1KihQ&amp;id=bXoqeH8eP" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>AI-generated code: The gift that keeps on giving… technical debt</strong>: Teams are finding that AI-generated code, while compiling initially, often lacks product context and creates architectural nightmares, especially with state management. One team even reported that refactoring AI-generated SwiftUI code took longer than building it from scratch. So, while AI might churn out lines of code at lightning speed, it seems we’re still stuck cleaning up its mess, proving that a fast mess is still a mess. <a href="https://app.daily.dev/feed-by-ids?id=6XeVWOBXO&amp;id=STWE5tRXe&amp;id=UAphs9F5B&amp;id=euzGVQGzQ&amp;id=iukrPGTg6&amp;id=m6P4U0J0b&amp;id=wCQfiCXDa" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Anthropic’s Claude Co-work: A ‘taste of AGI’ with a side of broken UX</strong>: Anthropic’s new Claude Co-work desktop app promises to bring AI superpowers to the masses, but early reports suggest it’s more of a ‘taste of frustration.’ While it can automate tasks in a sandboxed Ubuntu VM, users are battling broken authentication, UI glitches, and confusing permission models. It seems even a ‘powerful’ AI can’t fix a bad user experience, proving that sometimes, the ‘taste of AGI’ just tastes like a beta release. <a href="https://app.daily.dev/posts/OemgIV2rY" target="_blank" rel="noopener nofollow">Read more</a></li>
</ul>
