AI Development & Implementation
Lessons from building and testing AI agents and LLM applications in practice. I cover design, implementation, operations, and safety from a hands-on perspective.
Articles
What GPT-5.4, the Codex App, and Codex Security Reveal About OpenAI's Direction
Looking at GPT-5.4, the Codex app, and Codex Security from the week of March 2, 2026 together, this article explains how OpenAI appears to be moving beyond conversational and coding AI toward supporting a wider range of work.
Read moreWhat Is OpenAI Codex Security? I Tried It and Was Impressed by How Naturally It Leads to a Fix PR
I tested OpenAI Codex Security on a real GitHub repository and reviewed the full flow from security scanning to fix PR creation. This post covers a concrete SSRF finding and why the experience feels promising for AI-driven development.
Read moreWhat Is Harness Engineering? Designing the Rails That Keep AI Agents Stable
This post explains harness engineering as the outer structure that keeps AI agents stable during long-running, multi-step work. Based on practice, it covers initial alignment, design documents, checklists, handoff notes, review, and the rails around AI-assisted execution.
Read moreHow AI Agents Work: Models, Harnesses, Context, and Tools
This article explains how AI agents work and why similar requests can produce different results, using four elements: models, harnesses, context, and tools.
Read more