🍒
Cherry Servers
We give Developers full control and flexibility to build cloud-native applications on a powerful Cloud Platform.
Hello Builder,
This week’s newsletter highlights how Web3 and AI systems are becoming more structurally embedded in existing infrastructure: BVNK’s direct SEPA access and Ripple’s FCA-regulated services reflect crypto’s continued integration into traditional payments, while decentralized AI initiatives like Tempo’s partnership with DeepNode point to capital and compute shifting toward more open, programmable models.
Let's dive into the topics ⬇️:
🌐 Web3
BVNK is now a direct SEPA participant via the Bank of Lithuania’s CENTROlink, enabling 24/7 instant euro payments and seamless EUR–stablecoin transfers on one platform.
The project aims to broaden crypto adoption by tying AI and Web3 together: users keep control of their assets and interact with AI services without relying on a single company’s servers or permissions.
Solana’s first 2026 ecosystem update emphasizes that, despite flat crypto prices, development, user adoption, and institutional interest in Solana are all growing.
Ripple has secured FCA approval in the UK to offer regulated B2B crypto-related payment and e-money services, within a limited scope, focused on infrastructure rather than unrestricted retail crypto trading.
Attackers are impersonating Web3 employers and tricking candidates into installing fake “interview” tools or cloning malicious repos, leading to stolen wallet keys, credentials, and compromised dev environments. Any interview requiring unvetted software or code should be treated as a high-risk red flag.
📘 AI/ML
Anthropic is locking down how Claude can be used by third-party tools to prevent abuse of cheap plans for automation or training. This limits what developers can build, pushes power users to the paid API, and shows how easily closed AI platforms can change the rules for everyone.
Recommended by LinkedIn
AI projects don’t usually fail because of the models, but because the APIs around them aren’t built for autonomous agents. As teams move from demos to production, the real challenge is making APIs consistent, machine-readable, well-governed, and observable end to end.
Orchestral presents a simpler approach to building LLM and agent workflows, emphasizing reproducibility, easier debugging, and model-agnostic design. Positioned against LangChain’s growing complexity, the framework promotes clean, testable architectures over fragile “spaghetti agents.”
Delethink is a reinforcement learning method that teaches language models to periodically delete older reasoning tokens so they always think within a capped context window, while still benefiting from very long chains of thought overall.
🔧 DevOps
The idea is that simply “prompting better” isn’t enough. AI becomes reliable only when it’s used within a structured engineering process that includes clear design choices, validation, and review, thereby strengthening an existing workflow and producing production-ready results.
A Kubernetes 1.35 update changes how CSI drivers receive service account credentials, moving them out of volume_context and into the CSI secrets mechanism. This reduces the risk of accidental exposure, aligns better with the CSI spec, and requires CSI driver maintainers to opt in.
Istio acts as a control plane over Envoy sidecars, translating high‑level resources like VirtualService, DestinationRule, and AuthorizationPolicy into Envoy config, while EnvoyFilter is used when lower‑level control is needed.
The article emphasizes that most incidents come from ordinary tools misconfigured or underprotected, not novel zero-days, and that scale turns small oversights into major incidents.
💡 Did you know?
A step‑by‑step tutorial explaining how to deploy an OpenShift 4.19.11 cluster on three bare‑metal Cherry Servers using Red Hat’s Assisted Installer, static VLAN networking, and a Discovery ISO workflow.
