Investigations

Blade

TypeScript February 2026

An AI-assisted development methodology framework exploring structured approaches to building software with AI pair programming.

Blade is less about code and more about methodology — it's my attempt to formalise how I work with AI tools when building software. Rather than just chatting with an AI and hoping for the best, Blade defines structured workflows for different development tasks.

The framework covers the full development cycle: requirements gathering, architecture decisions, implementation patterns, testing strategies, and code review. Each stage has specific prompts, expected outputs, and quality criteria.

What I found most valuable was documenting the patterns that actually work versus the ones that sound good but produce inconsistent results. For example, giving the AI small, focused tasks with clear acceptance criteria works far better than dumping an entire feature specification.

The methodology has been battle-tested across most of my recent projects — you can see its influence in the BDD testing approach, hexagonal architectures, and incremental development diaries that show up throughout my portfolio.

What this project demonstrates

Structured AI-assisted development methodology
Covers requirements through code review
Specific prompts and quality criteria per stage
Battle-tested across multiple portfolio projects
Patterns for effective AI pair programming
Incremental development diary approach

Technologies

AIMethodologyDevelopment ProcessPair Programming

© 2026 Jonathan Leahy · v0.8.1-31-g196fa14