Writing
Long-form notes on AI monetization, AI-assisted engineering, and side projects.
Hands-On with Claude Code + Ralph Loop
A structured Plan → Generate → Execute → Reflect → Refine loop that turns AI from a one-shot generator into a collaborative iterative engineer.
An AI-Native Architecture Control System
We didn't add documentation rules — we built context governance for LLM collaboration: file contracts, module registries, and entropy control at AI velocity.
Spec-Driven Vibe Coding vs. Claude Code Plan Mode
Two AI coding workflows that look similar but optimize for different things — iteration speed under constraints vs. reliability under complexity.
Most Developers Are Using AI Wrong
The real leverage isn't faster typing — it's using Claude Code + Plan with Files as an architectural thinking engine before you generate code.
How AI Changes SaaS Pricing
Traditional seat-based models break down when intelligence becomes the product. Here's what's replacing them — and how to design pricing before your billing stack becomes the bottleneck.
Consumption Billing is Eating SaaS
Usage-based models aren't new, but AI workloads are accelerating the shift from subscriptions to consumption — and exposing gaps in legacy billing stacks.
Building Better Monetization Systems
Lessons from enterprise billing infrastructure — charge models, rating pipelines, observability, and why flexibility compounds over years.
Designing Pricing for Agents
Agent-native products need agent-native pricing — per-task, per-outcome, and hybrid models for software that acts autonomously.
A Primer on Prompt Engineering
Roles, parameters, and the building blocks of reliable LLM prompts — from system context to stop sequences.
Building a Hiking Route Intensity Calculator
Why I built a tool to unify incompatible difficulty formulas from clubs and social media posts.
How Great Engineers Ask for Help
A pre-escalation checklist and how to ask questions that build trust instead of signaling defeat.
Ramping Up on a New Codebase
A phased approach: map the system first, know where things live, then dive into code.