> I'm William. Here, I archive my observations on software architecture and AI-native systems, blending corporate leadership experience with a quantitative mindset.
A systems-level analysis of bi-temporal data architecture for trustworthy AI, focused on accuracy, latency, provenance, and operational trade-offs.
A system-architecture reading of AI skills, separating prompt text, workflow logic, and MCP-based tool execution so teams can reason about latency, control, and blast radius.
OpenClaw hype vs. reality: autonomous agents as glue layers, Copilot vs. mercenary metaphors, and why full local access or blind VPS quarantine both miss a sound architecture.