Code, Config, Comprehension
Three features shipped in a week, and each broke at a different checkpoint — a reminder that 'done' only ever certifies the layer you're standing on.
Read articleBlog
Ideas and lessons from delivery work.
What Clean Architecture Bought Me: a Firebase SQL Connect Trial
Two years after putting repository interfaces between the domain and Firestore, I wanted to swap the database. The Saturday-morning trial that followed turned on a single GraphQL annotation no one promised would work.
Read articlePrompt Harder Isn't a Strategy
Why the most confident advice in AI right now — 'just write a better prompt' — is a category error, and what to do instead.
Read articleThe Run I Didn't Take
On the cognitive cost of orchestrating six AI agents at once — and why the answer isn't fewer agents.
Read articleThree AI Tools, One Recommendation, No Clarifying Questions
What I learned asking Claude, Gemini, and Cursor to architect the same new mobile project — and why the tools that agreed with each other weren't necessarily right.
Read articleSplitting the Work: What AI Actually Automated for Me
On learning that the best use of AI isn't making it do the whole job — it's finding the chunk it can actually handle.
Read articleContext Is a Documentation Problem
Why AI coding agents hallucinate less when your codebase explains itself — and what to write to get there.
Read articleWhy AI-Enhanced Engineering Works Best With Strong Architecture
How architecture and practical engineering discipline turn AI tooling into sustainable delivery gains.
Read article