Stop fighting context decay. Part 5 introduces the 'Self-Updating' repository: using Context Patches to ensure your AI’s understanding evolves at the same rate as your implementation.
A collection of insights on software development
Stop fighting context decay. Part 5 introduces the 'Self-Updating' repository: using Context Patches to ensure your AI’s understanding evolves at the same rate as your implementation.
A Jira ticket shouldn't be a 'blind prompt.' Discover the Jira-to-Code workflow: a structured process where the AI audits requirements against your standards and maintains a 'Save Game' state for long-running tasks.
An AI 'expert' is only as good as the context it’s given. Part 3 dives into the technical core: defining your architectural blueprint into version-controlled standards that act as the AI’s compass and quality filter.
An AI that writes code without a plan is a technical debt generator. In Part 2, we explore Operational Roles — splitting AI into Architect and Developer modes to ensure every feature is designed, diagrammed, and validated before a single line is written.
Tired of re-explaining your tech stack to an AI every hour? Discover how to build a repository “Memory Centre” — a version-controlled source of truth that helps your AI respect your domain, stack, and architectural boundaries by default.
Part of a System Architect’s job is to choose tools that balance reliability, maintainability, and speed. In this post, I lay out the stack I’ll use across the blog — Terraform for IaC, Ansible for configuration, and Eraser.io → Mermaid.js for design — and the eight-question Systems Architect filter I’ll apply to every project.
What does it take to transition from a Senior Software Engineer to a Systems Architect in the AI era.