Sonnet 5 vs. Opus 4.8: When the Cheaper Model Costs More
Sonnet 5 is cheaper per token than Opus 4.8, yet in a real Claude Code refactor it cost nearly twice as much. Why agent behavior, not price, decides cost.
Writing
My Field Notes
Lessons from the day-to-day of building and leading: architecture, AI-assisted engineering, and the judgment a team runs on.
Cycle-Driven Engineering: The Organizational Architecture for AI-Assisted Engineering
The Constraint Layer is the organizational architecture of Cycle-Driven Engineering: decision ownership, boundaries, and reusable assets for AI-assisted delivery.
Why Most Code Reviews Should Never Happen
In the AI-assisted engineering era, reviewing code often costs more than writing it. How Cycle-Driven Engineering makes most code reviews unnecessary.
Fable 5 vs. Opus 4.8 – No Hype, Just Facts
Anthropic's Fable 5 costs double Opus 4.8 and arrives wrapped in hype. I put both to work head-to-head on the same workflow to see if the price is justified.
Why AI-Assisted Engineering Fails: The Vibe-Coder's Mindset
AI-assisted engineering fails when capable teams bring a vibe-coder's mindset to it. The difference is ownership, scaled from the engineer to the whole organisation.
How I Know My System Is Healthy, And What It Costs To Keep It That Way
Cost-to-serve, scaling, reliability, throughput. A CTO's measurement framework for any software product, with six months of real production data.
Claude Code Skills vs Sub-Agents (and Rules): How My Agent Learned My Business
How I use Claude Code skills, sub-agents, and rules, and when to choose each, so my AI coding agent understands my business concepts.
Cycle-Driven Engineering
Accelerating execution inside systems built for slow feedback does not work. It amplifies inconsistencies. Cycle-Driven Engineering is the answer.
Get My Field Notes in Your Inbox
Lessons from the day-to-day of building and leading, from the inside, as I build Atherio in real time. Free, no spam.