<h2>Must know</h2>
<ul>
<li><strong>JetBrains IDEs get native OpenAI Codex integration</strong>: JetBrains has integrated OpenAI Codex, powered by GPT-5.2-Codex, directly into its IDEs like IntelliJ, PyCharm, WebStorm, and Rider, starting with version 2025.3. This allows developers to use AI for refactoring, code migrations, and security vulnerability detection with adjustable autonomy levels and ‘reasoning budgets.’ Authentication options include JetBrains AI subscription, ChatGPT account, or an OpenAI API key, as reported by JetBrains. This is a significant step for AI-assisted development in a widely used ecosystem. <a href="https://app.daily.dev/posts/uAvdGTErT" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Waymo and Tesla expand unsupervised robotaxi services</strong>: Waymo launched its public robotaxi service in Miami, covering 60 square miles, with plans to expand to Miami International Airport and target 1 million trips per week by year-end. Meanwhile, Tesla began offering unsupervised robotaxi rides in Austin, gradually increasing the number of vehicles without human safety drivers. Both expansions come amidst ongoing federal safety investigations and reported traffic incidents, including Waymo vehicles passing stopped school buses. <a href="https://app.daily.dev/feed-by-ids?id=dBwgodiTL&amp;id=rPrfHRuqw" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Google DeepMind acquires Hume AI team for emotion-detecting voice AI</strong>: Google DeepMind has brought in Hume AI’s CEO and key engineers through a licensing agreement, signaling a major push into emotion-detecting voice AI. Hume AI, which reportedly anticipates $100 million in revenue this year, raised $80 million in funding. This move, structured as a licensing deal to potentially avoid regulatory scrutiny, highlights the increasing focus of tech giants like Google, OpenAI, and Meta on voice as the next frontier for AI interfaces. <a href="https://app.daily.dev/posts/Rb3T2e82m" target="_blank" rel="noopener nofollow">Read more</a></li>
</ul>
<h2>Good to know</h2>
<ul>
<li><strong>Feast integrates with PyTorch for enhanced feature management</strong>: Feast, an open-source feature store, has joined the PyTorch Ecosystem to provide a unified API for consistent feature management across AI model training and deployment, aiming to eliminate training-serving skew in production systems. <a href="https://app.daily.dev/posts/dPG1hvjH4" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Inferact raises $150M to commercialize vLLM inference tool</strong>: Inferact, a new startup, secured $150 million in seed funding at an $800 million valuation to commercialize vLLM, an open-source inference optimization tool already adopted by Amazon, highlighting growing investor interest in AI inference technologies. <a href="https://app.daily.dev/posts/qqbAFhp2s" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Microsoft launches VibeVoice-ASR for long-form audio speech-to-text</strong>: Microsoft released VibeVoice-ASR, a unified speech-to-text model capable of processing up to 60-minute audio files in a single pass, offering automatic speech recognition, speaker diarization, and timestamping for meeting and conversational scenarios. <a href="https://app.daily.dev/posts/zq6H3l6U4" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Rust 1.93.0 updates musl libc and stabilizes APIs</strong>: Rust 1.93.0 was released, updating the bundled musl libc to version 1.2.5 for improved DNS resolution in static Linux binaries and stabilizing 23 APIs, including <code>MaybeUninit</code> slice operations and unchecked arithmetic. <a href="https://app.daily.dev/feed-by-ids?id=wF6DW7Hp4&amp;id=O9g10PE7Q" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>ClickHouse launches managed Postgres service with native integration</strong>: ClickHouse introduced a managed Postgres service, promising up to 10X faster performance for disk-bound workloads using NVMe storage, with native integration via CDC and the <code>pg_clickhouse</code> extension, and all integration tools released as open source. <a href="https://app.daily.dev/posts/WHTgaLabX" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>GitHub Actions offers 1 vCPU Linux runners</strong>: GitHub Actions now provides generally available 1 vCPU Linux runners (<code>ubuntu-slim</code>) as a lower-cost option for lightweight automation tasks, offering 5GB RAM and 15-minute execution limits for jobs like auto-labeling issues and basic compilation. <a href="https://app.daily.dev/posts/2Mmk9fDWb" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Cerbos Hub now available for on-premise deployment</strong>: Cerbos Hub has released support for on-premise deployment, allowing organizations with strict data residency or air-gapped requirements to run its centralized authorization infrastructure, including policy management and GitOps workflows, within their own environments. <a href="https://app.daily.dev/posts/I7gHtDX2Z" target="_blank" rel="noopener nofollow">Read more</a></li>
</ul>
<h2>Bullshit police</h2>
<ul>
<li><strong>AI agents for IT operations: what could go wrong?</strong>: Apparently, AI agents are the new ‘cowboy sysadmins’ for IT operations, introducing ‘non-deterministic actions’ and increasing risk. The article, from some unnamed source, warns against letting AI directly modify live production systems, suggesting they stick to generating playbooks and Dockerfiles. Because, you know, we definitely need more playbooks and Dockerfiles that still require ‘human review, testing, and reproducibility’ before they touch anything important. Sounds like a lot of extra steps for something that’s supposed to be autonomous. <a href="https://app.daily.dev/posts/U8PYVjc73" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>AI-driven QA analytics: the magic bullet for executives</strong>: Testlio launched LeoInsights, an AI-based QA analysis platform trained on 13 years of their own data, promising ‘executive-level summaries’ and a ‘value calculator’ to measure efficiency. Because what executives really need is an AI to tell them how much money they’re saving, rather than, say, actual independent verification of product claims or competitive analysis. It’s all about converting ‘software quality metrics into actionable business insights,’ which I’m sure will be totally unbiased and not at all promotional. <a href="https://app.daily.dev/posts/xZI6Dgtwy" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>AI agents for vulnerability assessment: trust us, it’s over 90% accurate</strong>: ZEST Security introduced AI Sweeper Agents that claim to identify which cybersecurity vulnerabilities are ‘truly exploitable’ in your environment, dismissing 11 million vulnerabilities in six months for early adopters. They say ‘over 90%’ of high and critical vulnerabilities aren’t actually exploitable. This sounds fantastic, until you realize all these impressive numbers come directly from ZEST Security itself, with no external validation. So, we’re just supposed to trust that their AI knows best, even when it contradicts traditional security assessments? What could possibly go wrong? <a href="https://app.daily.dev/posts/LDTl5G8Jm" target="_blank" rel="noopener nofollow">Read more</a></li>
</ul>
