Blog
Field notes, post-mortems, and technical logs written by a developer working on the surface.
TL;DR: Technical journals detailing UI engineering workflows, Next.js architecture, legacy software migrations, and automation strategies.
AI Engineering • 9 min read
Published: June 14, 2026
A step-by-step guide to setting up recursive agent fanout, parallel greenfield generation, and autonomous autoresearch loops using GPT-5.5 xhigh in the Oh My Pi harness.
AI Workflows • 8 min read
Published: June 7, 2026
A practical build log from creating MoolaMochi transactional emails with Codex, Paper MCP, hosted assets, and generated artwork that shipped into real artboards instead of placeholder boxes.
AI Workflows • 7 min read
Published: June 7, 2026
A practical field log from rebuilding the MoolaMochi icon system: removing white backgrounds, regenerating damaged assets, and using Codex plus GPT Image Gen 2 to keep every icon in the same visual family.
AI Workflows • 6 min read
Published: April 27, 2026
AI agents are incredibly powerful, but asking them to invent a premium design from scratch usually leads to average results. Here is why you should separate design from execution, generating high-fidelity UI references first and coding them section-by-section.
AI Workflows • 7 min read
Published: May 1, 2026
A step-by-step example of the Image-to-Code workflow: generating section comps in ChatGPT, extracting sub-assets, removing backgrounds, and executing layout code with Codex.
AI Engineering • 6 min read
Published: June 8, 2026
Why does the same AI model work like a genius for one developer and fail constantly for another? The answer lies in separating the harness, the model, and the serving infrastructure.