The Development Method
How to use AI to build software and design systems through spec-driven, ratcheting vertical slices, clear ownership boundaries, and repeatable verification — so progress stays forward-only, auditable, and grounded in real working applications.
THE ARGUMENT
THREE CLAIMSMove in verified slices
Development advances in thin, end-to-end vertical slices — each traced to a requirement and proven three ways before the next begins. Forward-only, like a ratchet.
Keep ownership clear
A clean boundary between coding and operations. The developer + local AI build inside the repo; a DevOps layer handles continuity, deployment, and diagnosis around the running system.
AI as guided collaborator
AI is strongest given references, examples, and patterns — generating direction, exploring alternatives, and converging under human judgment. Not an autonomous replacement for it.
THE DECK
OPEN LAUNCHER →Recommended flow — nine slides building a coherent argument.
Sample Pages
A hub showing the scope of the material and the live sample projects it draws on.
The Problem
The central issue wasn't only code — it was ownership around the code, and the danger of mixing app work, environment, deployment, and diagnosis in one loop.
What It Solves
The failure modes the method prevents: context loss, infrastructure improvisation, execution drift, and product-delivery drift.
The Development Method
Spec-driven, ratcheting vertical slices — verified three ways. Scope, build, verify, deployable state, and ratchet forward.
Applications & Operating Model
The method inside a larger operating model with live surfaces — the Simon, Simon2, and Pubert roles as segmented responsibility and continuity.
My Setup
The actual operating environment: GitHub SSH access, a live web-AI surface reachable from Discord, flexible tool choice, and permissioned autonomy with recovery in place.
Remote Agents
Continuity plus delegation — leave context, connect remotely, let bounded agents execute, and preserve visibility for review and recovery.
AI Design Generation
The shift from software process to design process — AI works best given references, examples, and code patterns instead of vague prompts.
The Iteration Mindset
AI as an iterative loop — generate, evaluate, tweak, and converge — with human judgment as the controlling force.
LIVE SAMPLE PROJECTS
REAL WORKING SURFACESThese live surfaces are what the method produced — the deck uses them to show the workflow in practice, not in the abstract.
Spec-driven · ratcheting · vertical-slice · three-way-verified. Small forward-only steps turned documents into a working, versioned product — with the audit trail built in.
LAUNCH THE PRESENTATION play_arrow