<h2>Must know</h2>
<ul>
<li><strong>AI agents are transforming software development and operations</strong>: AI agents are rapidly changing how software is built and operated, with companies like Stripe using ‘minions’ to autonomously merge over 1,000 pull requests weekly. Microsoft is rolling out ‘agentic cloud operations’ for optimizing cloud environments, and Cursor’s long-running agents are completing complex projects in hours. This shift means developers are becoming orchestrators, defining intent and managing AI, rather than writing every line of code, as highlighted by GitHub co-founder Scott Chacon. <a href="https://app.daily.dev/feed-by-ids?id=HK49rZyW0&amp;id=HK49rZyW0&amp;id=XwFeQESAO&amp;id=a6tJErMuD&amp;id=KgDUr7AXX&amp;id=POybmTTi0" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>AI infrastructure faces critical bottlenecks and market shifts</strong>: The demand for AI is driving massive infrastructure investments, with India aiming for over $200 billion by 2028 and Meta signing a $6 billion fiber optic deal with Corning to address the ‘networking wall’ for large AI models. NVIDIA’s Blackwell Ultra platform promises up to 50x higher throughput and 35x lower cost for agentic AI workloads, while specialized ‘neoclouds’ are emerging to offer GPU-first computing at less than half the cost of hyperscalers. This intense competition for resources is also causing memory price surges and hardware shortages across the tech industry. <a href="https://app.daily.dev/feed-by-ids?id=hAmXnwmkV&amp;id=iCAaBNNNy&amp;id=Hamza3qTP&amp;id=Hamza3qTP&amp;id=2G5oHg0oJ&amp;id=yYXkpJGbb&amp;id=XLkxKsrjj&amp;id=aAvHNcO4S&amp;id=w4htkmXsV" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>AI agents introduce new security and governance challenges</strong>: The rapid adoption of AI agents is creating significant security vulnerabilities, with prompt injection attacks evolving into ‘promptware’—multi-stage malware campaigns that exploit AI’s lack of architectural boundaries. OpenClaw, a popular AI agent, has been criticized as a ‘self-inflicted root kit’ due to its broad system access and lack of secure-by-default configurations. GitGuardian recently secured $50 million to address this ‘non-human identities crisis,’ emphasizing the need for robust governance frameworks and specialized security for autonomous AI systems. <a href="https://app.daily.dev/feed-by-ids?id=zpTLornb8&amp;id=y2ffWaFo0&amp;id=Mx18Wv9Wt&amp;id=jYkCxtUoh&amp;id=l9nJMN2FZ&amp;id=mmQEWnSC4&amp;id=UhlBeqTGk&amp;id=YqMdmrbZu&amp;id=MZmobK75f&amp;id=1X0JlaofS" target="_blank" rel="noopener nofollow">Read more</a></li>
</ul>
<h2>Good to know</h2>
<ul>
<li><strong>New AI models and tools for developers</strong>: MiniMax launched M2.5, a 230B parameter model claiming to rival Claude Opus 4.6 at one-tenth the cost, excelling in agentic coding and multi-file refactoring. Google released Gemini 3 Flash, combining frontier-level intelligence with high speed and low cost, and DeepMind introduced Gemma Scope 2, an open-source toolkit for AI interpretability. Meanwhile, Kiro IDE now supports open-weight models like DeepSeek v3.2 and Qwen3 Coder Next, and Heroku launched sandboxes for AI agents to run untrusted code securely. <a href="https://app.daily.dev/feed-by-ids?id=VQt2ETdNG&amp;id=jGb0yTdHY&amp;id=KpugFpJBL&amp;id=XSXe8Y92K&amp;id=Ke3nu27Xr&amp;id=bzMzHxxko" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>PostgreSQL continues to scale and innovate</strong>: OpenAI scaled a single-primary PostgreSQL instance to handle millions of queries per second for ChatGPT’s 800 million users, leveraging read replicas and connection pooling. Supabase acquired Hydra, integrating pg_duckdb to accelerate analytics queries by over 600x, and VillageSQL launched an open-source MySQL fork aiming for PostgreSQL-style extensibility. ClickHouse open-sourced pg_stat_ch, a PostgreSQL extension for streaming query execution events for analysis. <a href="https://app.daily.dev/feed-by-ids?id=0QBTpoQPy&amp;id=wcYLdtADY&amp;id=Yc1LRcEqx&amp;id=Yc1LRcEqx&amp;id=VSfuUpVNX&amp;id=VSfuUpVNX&amp;id=WTP9AqFIZ&amp;id=WTP9AqFIZ&amp;id=aVu4MLQ4Y&amp;id=aVu4MLQ4Y" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>AI is reshaping software development workflows</strong>: AI coding assistants are shifting the bottleneck from writing code to specifying desired outcomes and verifying results, making Test-Driven Development (TDD) and Behavior-Driven Development (BDD) critical. Continue.dev launched a free VS Code extension for automated code reviews, and GitHub Copilot Testing for .NET is now generally available for AI-powered unit test generation. However, AI-generated code often produces bloated, insecure output, requiring experienced developers to review and refactor. <a href="https://app.daily.dev/feed-by-ids?id=ZrBLHCGzU&amp;id=vCiFcLhDP&amp;id=sJ5P2kSW8&amp;id=gvdBYayNu" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>New standards and protocols for AI agents emerge</strong>: Google announced the A2A (Agent-to-Agent) protocol with over 50 tech partners to enable direct AI agent communication, complementing Anthropic’s MCP standard. W3C proposed WebMCP to allow web developers to expose structured tools directly to in-browser AI agents, aiming to replace inefficient web scraping. OpenAI published the architecture of Codex App Server, a bidirectional protocol to unify all Codex coding agent experiences through a single API. <a href="https://app.daily.dev/feed-by-ids?id=vb6x1hImO&amp;id=wj4NMewlB&amp;id=xHgjVo7OS&amp;id=0nYgeKiTK&amp;id=zlD0DdN1A&amp;id=V16f6xKqK" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Cloud providers enhance AI and virtualization capabilities</strong>: AWS announced support for nested virtualization on virtual EC2 instances, allowing customers to run KVM or Hyper-V inside. AWS also launched new M8azn and Hpc8a EC2 instances powered by 5th Gen AMD EPYC processors, offering significant performance gains for compute-intensive workloads. Amazon Bedrock expanded support for six new open-weight models, including DeepSeek V3.2 and Qwen3 Coder Next, with OpenAI API compatibility. <a href="https://app.daily.dev/feed-by-ids?id=hFJsILiy4&amp;id=dNY05UeZN&amp;id=fFNzT4nn1&amp;id=alSlJC9VU" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>AI is reshaping the gaming industry and hardware supply</strong>: ByteDance’s Seedance 2.0, an AI video generator, is facing Hollywood IP backlash for reproducing copyrighted characters and actors. Meanwhile, the ‘AI boom’ is contributing to hardware shortages, potentially delaying Sony’s next-gen PlayStation and driving up prices for Nintendo’s Switch 2. Krafton, maker of PUBG, is outlining ambitious AI plans, suggesting its in-game AI could transfer to humanoid robotics. <a href="https://app.daily.dev/feed-by-ids?id=H0AMOrypO&amp;id=VHhfvW7O6&amp;id=bKEf8FvYb&amp;id=SBvqFuIKt" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Java ecosystem adapts to AI and licensing changes</strong>: Java’s adoption for AI development increased to 62% year-over-year, with JavaML and Deep Java Library being popular choices for production AI workloads. However, 92% of Java users are concerned about Oracle’s licensing changes, driving 81% to migrate to open-source alternatives like Bellsoft Liberica. Jakarta EE 12 Milestone 2 introduced a new Jakarta Agentic AI specification to standardize AI agent development on Jakarta EE runtimes. <a href="https://app.daily.dev/feed-by-ids?id=TvY0j5h5O&amp;id=LDTrFcHSl&amp;id=n4gtBvbDs&amp;id=ItqM59Igw&amp;id=ItqM59Igw&amp;id=0V8Ax2TZk&amp;id=QRBfTNO1G&amp;id=zhW5HcoVn&amp;id=Q59oI3NLH" target="_blank" rel="noopener nofollow">Read more</a></li>
</ul>
<h2>Bullshit police</h2>
<ul>
<li><strong>AI’s impact on software development: the great developer identity crisis</strong>: A Berkeley Haas study found AI tools intensify workload and cause burnout, while a veteran programmer feels a loss of identity due to AI-generated code. Anthropic’s CEO predicted AI would write ‘essentially all code’ within 12 months, and some developers claim they’re already there. So, AI is making us more productive, but also more miserable, less skilled, and questioning our very existence. Sounds like a win-win for everyone, right? <a href="https://app.daily.dev/feed-by-ids?id=2nGSglCla&amp;id=3BYKTGamX&amp;id=xfda4p1KO&amp;id=xjDRiBzgJ&amp;id=hewIkht73&amp;id=NY3oBIud2" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>AI hype cycle: trust erosion and ROI challenges</strong>: Public trust in AI is collapsing, falling to 32% in the US, as 95% of enterprise generative AI pilots yield no measurable ROI despite $40 billion in spending. LLMs hallucinate legal citations 17-33% of the time, yet governments use them for critical decisions. Meanwhile, SEC enforcement actions against ‘AI washing’ increased 100%. So, we’re spending billions on AI that doesn’t work, nobody trusts, and is getting companies sued. But hey, at least the stock market is having a good time, right? <a href="https://app.daily.dev/feed-by-ids?id=4eW9DctUo&amp;id=PryOnuBPQ&amp;id=AN7GCbcU0&amp;id=BEYbuELEG&amp;id=C1KISrsmg&amp;id=C7OGVNLnN" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>AI is destroying open source, and it’s not even good yet</strong>: Open-source maintainers are being overwhelmed by low-quality, AI-generated pull requests, with the curl project dropping bug bounties due to AI spam. Godot co-founder Remi Verschelde reports maintainers are demoralized by scrutinizing every new contribution for AI-generated errors. GitHub now even allows repositories to disable pull requests entirely. So, AI is so ‘helpful’ it’s making open source unusable, forcing maintainers to either drown in garbage or shut down contributions. Progress! <a href="https://app.daily.dev/feed-by-ids?id=p4VqhMMFm&amp;id=FbzHThFRg&amp;id=FbzHThFRg&amp;id=GITxcoUi5&amp;id=GAkIuPiut&amp;id=3TeLmhBBy" target="_blank" rel="noopener nofollow">Read more</a></li>
</ul>
