Ai

Architecture discussions go sideways because everyone is reasoning at internally inconsistent altitudes, detailed where they’re fluent and hand-wavy where they’re not. C4 Level 2, the container diagram, is the altitude where abstraction stops lying: real processes, real protocols, real failure modes. That’s why it aligns humans, why the same framing works on LLM agents, and why fluency at this altitude is now table stakes for engineering in the agentic age.

The AI coding tool landscape is changing fast, but the underlying models share the same strengths and limitations. I ran a lunch-and-learn focused on building the mental model that makes you effective regardless of which tool you’re using — how to structure problems for an LLM, why planning before execution matters more than ever, and why clarity is now your most valuable engineering skill. We tested it by building a working app live in under 30 minutes.

I’ve started using CLI-based coding agents for general planning and documentation because local file context is easier to manage than web interfaces. The only downside is sharing the Markdown output with others. This post covers the workflow and a small tool I wrote to convert that output into shareable PDFs.