Check Point Research

Flutter

GitHub Changelog

Hacker News

JetBrains

Joshua Morony

<h2>Must know</h2>
<ul>
<li><strong>OpenAI Codex and Claude Code now integrated into JetBrains IDEs</strong>: JetBrains IDEs, including IntelliJ, PyCharm, WebStorm, and Rider, now natively integrate OpenAI Codex (powered by GPT-5.2-Codex) and Claude Code. This allows developers to access AI agents for tasks like refactoring, code migrations, and security vulnerability detection directly within their IDEs, with adjustable autonomy levels and ‘reasoning budgets.’ JetBrains itself confirmed the integration, and users can authenticate via a JetBrains AI subscription, ChatGPT account, or OpenAI API key. This is a significant step towards embedding advanced AI directly into daily development workflows, as noted by JetBrains. <a href="https://app.daily.dev/feed-by-ids?id=uAvdGTErT&amp;id=IHws5ybw2" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>AI agents are changing how we manage code and infrastructure</strong>: The rise of AI agents operating at machine speed is forcing a re-evaluation of traditional version control and infrastructure management. Atomic Software has released a new version control system, based on Pijul’s patch theory, specifically designed for AI-generated code, addressing the validation and tracing bottlenecks that Git can’t handle. Additionally, the concept of ‘Agent Skills’ is emerging as an open standard to package AI instructions into reusable, reproducible workflows, treating prompts as version-controlled infrastructure. This shift, highlighted by articles from Flutter and Atomic Software, aims to bridge the gap between rapid AI code generation and the need for consistent, high-quality, and auditable development. <a href="https://app.daily.dev/feed-by-ids?id=xcY01cgn0&amp;id=m5IRqcNSP" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Microsoft launches unified speech-to-text model and AI-powered Notepad</strong>: Microsoft has 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. This model, according to Microsoft, is designed for meeting and conversational scenarios. Concurrently, Notepad received an update with AI streaming improvements for ‘Write, Rewrite, and Summarize’ functions, requiring a Microsoft account sign-in for these features. These updates, confirmed by Microsoft, signal a push to integrate AI more deeply into both core productivity tools and specialized services. <a href="https://app.daily.dev/feed-by-ids?id=zq6H3l6U4&amp;id=JTjjU1lE2" target="_blank" rel="noopener nofollow">Read more</a></li>
</ul>
<h2>Good to know</h2>
<ul>
<li><strong>GitHub releases multi-platform SDK for Copilot agent integration</strong>: GitHub has launched a multi-platform SDK for Python, TypeScript, Go, and .NET, enabling integration of Copilot’s agentic workflows, including orchestration, planning, and programmatic file edits, into various applications. <a href="https://app.daily.dev/feed-by-ids?id=GXXnGjqMp&amp;id=GuG7K2ryZ" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Datadog unifies RUM and APM data with automatic retention filters</strong>: Datadog introduced cross-product retention filters that automatically retain Application Performance Monitoring (APM) traces linked to Real User Monitoring (RUM) sessions, unifying frontend and backend observability data for improved troubleshooting. <a href="https://app.daily.dev/posts/fs5040Sq6" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Feast integrates with PyTorch ecosystem 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, addressing 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>Pulumi adds replacementTrigger for automated resource recreation</strong>: Pulumi v3.215.0 introduces <code>replacementTrigger</code>, allowing infrastructure-as-code resources to be automatically recreated when specified metadata changes, useful for cycling secrets or updating external dependencies. <a href="https://app.daily.dev/posts/ay0AhT4GQ" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Rust 1.93.0 improves musl libc and inline assembly handling</strong>: Rust 1.93.0 updates the bundled musl libc to 1.2.5, enhancing DNS resolution for static Linux binaries, and stabilizes 23 APIs, including improved handling of inline assembly macros. <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>Vercel Sandboxes introduces filesystem snapshot feature</strong>: Vercel Sandboxes now supports filesystem snapshots, allowing users to save the complete state of a running sandbox and launch new ones from these snapshots, streamlining CI/CD workflows and rapid prototyping. <a href="https://app.daily.dev/posts/Uv3Zx9otG" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>ClickHouse launches managed Postgres service with native integration</strong>: ClickHouse has released a managed Postgres service, promising up to 10X faster performance for disk-bound workloads and native integration with ClickHouse via CDC and pg_clickhouse, enabling a unified transactional and analytical data stack. <a href="https://app.daily.dev/posts/WHTgaLabX" target="_blank" rel="noopener nofollow">Read more</a></li>
</ul>
<h2>Bullshit police</h2>
<ul>
<li><strong>AI agents are not ready for prime time in the workplace</strong>: Mercor’s new Apex-Agents benchmark, designed to test AI models on ‘real-world white-collar tasks,’ found that leading models like Gemini 3 Flash and GPT-5.2 achieved a whopping 24% accuracy. Apparently, AI agents can’t reliably replace knowledge workers in complex, multi-tool tasks, especially when ‘multi-domain reasoning’ is involved. So, while the hype machine keeps churning, it seems your job is safe from our silicon overlords for a little while longer. Don’t worry, they’re still working on it, but maybe focus on actual productivity for now. <a href="https://app.daily.dev/posts/OsmrZ9UpM" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>eBay bans AI shopping agents, because who needs efficiency?</strong>: eBay, in a bold move to protect… something, has updated its User Agreement to explicitly ban third-party AI agents and chatbots from making purchases without permission. This is apparently a response to the ‘rise in agentic commerce tools’ and the terrifying prospect of ‘end-to-end purchasing flows that bypass human review.’ Because nothing says progress like forcing humans to click buttons when an AI could do it faster. I guess we’re not ready for a world where shopping is actually efficient, or maybe eBay just wants to make sure you still see all their ads. <a href="https://app.daily.dev/posts/eNz11CwvB" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Yann LeCun declares LLMs dead, launches new AI venture</strong>: Yann LeCun, the guy who helped invent modern AI, has launched AMI Labs in Paris, explicitly stating that large language models are fundamentally limited because they ‘lack understanding of the physical world, inability to reason, and inability to plan.’ His solution? ‘World models’ trained on video, audio, and sensor data to create ‘abstract representations of reality.’ So, while everyone else is still trying to make LLMs less hallucinatory, LeCun is off building something entirely different. Good luck with that, Yann, but I’m sure the next generation of AI hype will be just as grounded in reality as the last. <a href="https://app.daily.dev/feed-by-ids?id=bsXOFS9xq&amp;id=wOdmxGV2q" target="_blank" rel="noopener nofollow">Read more</a></li>
</ul>
