<h2>Must know</h2>
<ul>
<li><strong>MCP servers are evolving with better security and developer tools</strong>: The Model Context Protocol (MCP) is gaining traction, with new implementations like Context7’s OAuth 2.1 integration and YepCode’s developer-first platform. These developments aim to standardize how AI agents interact with external tools and APIs, addressing critical aspects like authentication, security, and governance. This is crucial for building robust and secure AI agent architectures that can safely access internal systems and data, moving beyond manual API key setups and enabling more flexible, scalable deployments. <a href="https://app.daily.dev/feed-by-ids?id=Pa4anLLMe&amp;id=TTuEqrxV6&amp;id=fBtDP8BjG&amp;id=gSf3nPjbc" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Kubernetes Ingress NGINX is being deprecated, forcing migrations</strong>: The Kubernetes ecosystem is facing a significant shift as Ingress NGINX, used by 50% of cloud-native environments, will stop receiving security patches in March 2026 due to maintainer burnout. This necessitates migration to alternatives like Gateway API, Envoy Gateway, Traefik, or Cilium. This situation highlights the challenges of open-source sustainability and the need for platform teams to proactively manage critical infrastructure dependencies, especially in Kubernetes deployments. <a href="https://app.daily.dev/feed-by-ids?id=wK9Dxbori&amp;id=PptBbhnAu" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Upwind security raises big bucks for AI-driven cloud protection</strong>: Upwind Security, a cloud security platform, raised $250 million at a $1.5 billion valuation. Their “inside-out” approach uses internal signals like network requests and API traffic to prioritize real-time threats, contrasting with traditional agentless scanning. This funding will support the expansion of AI security capabilities and developer-focused misconfiguration prevention, indicating a growing market for advanced, AI-driven cloud security solutions. <a href="https://app.daily.dev/posts/LGgCBPA2D" target="_blank" rel="noopener nofollow">Read more</a></li>
</ul>
<h2>Good to know</h2>
<ul>
<li><strong>LLM observability slashes debugging time for AI agents</strong>: LLM observability is proving critical for debugging AI agents, with Datadog reducing debugging time from hours to minutes by using ground truth datasets and component-level evaluators. <a href="https://app.daily.dev/posts/RGWEV4Kqk" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Cloud Run gets NVIDIA GPUs for serverless AI workloads</strong>: Google Cloud Run now supports NVIDIA RTX PRO 6000 Blackwell GPUs, enabling serverless deployment of large AI models (up to 70B+ parameters) without cluster management. <a href="https://app.daily.dev/posts/Vh4SzEb3w" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Google Cloud Spanner becomes an AI context hub</strong>: Google Cloud Spanner has evolved into an intelligent context hub for AI applications, offering multi-model support, AI-powered operations, and unified analytics. <a href="https://app.daily.dev/posts/TRihFu0x6" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>OpenSearch boosts AI agent search with vector embeddings</strong>: OpenSearch is enabling search modernization beyond keywords through vector embeddings, hybrid search, and LLM-powered query intent understanding for AI agents. <a href="https://app.daily.dev/posts/GhJZtEl9V" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>New foundation models land on AWS SageMaker JumpStart</strong>: AWS has added DeepSeek OCR, MiniMax M2.1, and Qwen3-VL-8B-Instruct to SageMaker JumpStart, expanding foundation model options for document processing, coding, and multimodal reasoning. <a href="https://app.daily.dev/posts/wEnEYbP9z" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Google Cloud launches Axion ARM instances</strong>: Google Cloud’s N4A instances, powered by Axion ARM processors, are now generally available, offering cost-effective performance for web servers, microservices, and containerized applications. <a href="https://app.daily.dev/posts/JvBjVyZ5U" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>OpenEverest automates database management on Kubernetes</strong>: Percona announced OpenEverest, an open-source platform for automated database provisioning and management on Kubernetes, supporting MySQL, PostgreSQL, and MongoDB. <a href="https://app.daily.dev/posts/hNifCVJEN" target="_blank" rel="noopener nofollow">Read more</a></li>
</ul>
<h2>Bullshit police</h2>
<ul>
<li><strong>Tesla robotaxis crash more than humans, hides data</strong>: Apparently, Tesla’s robotaxi fleet in Austin has a crash rate three times worse than human drivers, and nine times worse for police-reported incidents. And to top it off, Tesla redacts all crash narratives in its NHTSA reports, unlike Waymo, which operates truly driverless vehicles and has a better safety record. Transparency? Who needs that when you have Elon’s promises? <a href="https://app.daily.dev/posts/M8WkrOIKM" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Open-source SwiftUI agent skill promises to fix all your coding woes</strong>: A new open-source SwiftUI Agent Skill promises to prevent common coding mistakes and reduce technical debt by embedding SwiftUI expertise into AI agents. Because, you know, AI is just <em>so good</em> at understanding context and preventing the kind of nuanced errors that human developers spend years learning to avoid. What could possibly go wrong? <a href="https://app.daily.dev/posts/LlEDXwb6h" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>AI-coded tool reviews AI-coded code, what could go wrong</strong>: GitHuman, a new local-first tool for reviewing AI-generated code, was entirely written by Claude Code, an AI model. And get this: every commit was human-reviewed using GitHuman itself! It’s a self-referential loop of AI-generated code reviewing AI-generated code, with a human somewhere in the middle, probably just nodding along. What a time to be alive, where the tools we use to review AI code are themselves AI-generated. Trust the process, I guess. <a href="https://app.daily.dev/posts/M78k9F0lW" target="_blank" rel="noopener nofollow">Read more</a></li>
</ul>
