🍒
Hello Builder,
This week’s wrap-up: Hyperliquid released its HIP-3 upgrade, making transactions faster, cheaper, and more secure by letting people trade without intermediaries. Coinbase applied for a national trust charter in the US, aiming for clearer regulation and stronger user trust. Claude Code now supports plugins, so developers can easily add tools for tasks like testing or code analysis. OpenAI teamed up with Broadcom to build its own AI chips, reducing dependence on Nvidia and gaining more control over costs and performance. Bybit integrated UBS’s uMint tokenized fund as trading collateral, further bridging traditional finance and Web3. And there’s more waiting inside the newsletter, so don’t miss the latest AI and DevOps stories we saved for you.
Let's dive 🚀 into the topics:
🌐 Web3
Bybit has integrated UBS’s uMint tokenized fund as trading collateral on its platform, marking a significant step in bridging traditional finance and Web3 innovation. This integration allows users to use the institutional-grade uMint money market fund, launched by UBS on Ethereum in 2024, as collateral for trading, enhancing capital efficiency and transparency for professional and institutional investors.
Coinbase is part of a growing trend, joining companies like Circle, Paxos, and Ripple in seeking federal charters as the U.S. regulatory environment shifts in favor of digital asset providers. This approach is expected to deliver operational advantages, greater trust, and more innovative digital finance products in the U.S. market.
The HIP-3 upgrade on Hyperliquid introduces permissionless perpetual futures markets powered by HYPE token staking. By implementing permissionless, secure markets and garnering strong adoption, Hyperliquid sets a new benchmark that may drive competing platforms to innovate and raise their own standards.
Democrats in the U.S. Senate have proposed a "restricted list" for DeFi protocols that could have major consequences for the sector, according to widespread reporting and industry reaction. The proposal seeks to expand the scope of regulation by designating certain decentralized finance platforms and applications for heightened scrutiny or outright restriction.
📘 AI/ML
Claude Code now supports plugins, allowing users to extend its capabilities with custom collections of slash commands, specialized agents, MCP servers, and workflow hooks that install with a single command.
The paper introduces ReasoningBank, a memory framework for large language model agents to leverage their accumulated task experiences by distilling generalizable reasoning strategies from both successes and failures. At test time, agents use these distilled memories to guide new interactions and update the memory with fresh insights, making the agents more capable over time.
OpenAI has partnered with Broadcom to develop its first custom artificial intelligence processors, marking a major shift in how OpenAI sources the computational infrastructure necessary for large-scale AI technology. This collaboration aims to reduce OpenAI's reliance on established chipmakers like Nvidia and AMD, granting it more control and flexibility over its hardware as AI workloads grow more demanding.
Recommended by LinkedIn
The study has raised alarm about the accessibility and seriousness of backdoor vulnerabilities in AI models trained on large datasets, pushing the field to reconsider how training data integrity is guaranteed and protected.
While the failures highlight unresolved reliability issues, especially with citations and report writing, Deloitte’s investment signals faith in ongoing improvement and the inevitability of AI in daily business operations for efficiency and scalability.
This certification demonstrates that Runpod's security controls and processes have undergone rigorous, independent auditing over an extended period, offering customers additional assurance that their data is managed in line with best practices for availability, confidentiality, and integrity.
🔧 DevOps
Adopting end-to-end encryption in Git services could meaningfully reduce risks associated with both external and insider threats, making source code management more secure in today's cloud-driven development landscape.
Kubernetes delivers immense value for enterprise Java by managing microservices complexity and automating many infrastructure concerns, but it introduces new challenges in configuration and operational overhead. The future lies in specialized, automated platforms that make K8s truly manageable for Java teams, so developers can focus on business logic rather than infrastructure management.
key updates for open source software, important ecosystem developments, and a significant hardware launch. The wrap-up covers the release of Ubuntu 25.10, new features in various open-source tools, the acquisition of Arduino by Qualcomm, new tech initiatives, and the launch of a high-end Linux laptop from System76.
MySQL deadlocks occur when two or more transactions wait for each other to release locks, causing all to be stuck until MySQL rolls one back. Most deadlocks are a normal risk in systems with high concurrency but frequent or repeating patterns indicate deeper structural or logic issues
💡 Did you know?
Kubernetes on bare metal offers flexibility, performance, and cost savings for technical teams. OpenShift on bare metal delivers enterprise integration, security, and operational efficiency—at a higher resource and licensing cost. The best choice depends on your organizational needs, expertise, compliance, and budget.