My AI agent tried to do a single line edit and got blocked by its own governance rules. That's not overhead — that's the system working. Here's what real AI throughput looks like when you stop babysitting and start governing.
Designing resilient cloud systems and practical AI workflows.
I build enterprise-grade platforms and focused AI integrations, then share the architecture decisions, constraints, and delivery patterns that work in production.
Find the thing that's slow, fragile, or tedious. Make it not that.
Featured
Latest Writing
I spent more hours building AIAgentMinder than I did on the 11 sprints where I used it to ship a real product. A quick look at what that means — and what a two-minute slash command told me about the state of my project.
How to route automated and real-time Claude requests through your Max subscription instead of paying per-call — with two production patterns: an n8n sub-workflow for async operations and an Express gateway for synchronous app integration.
AIAgentMinder started as a memory system for Claude Code with rules and planing mode. Then Anthropic shipped native memory. Then they shipped rules. Then they shipped plan mode. Here's what survived, what didn't, and what I learned about building tools on a platform that moves faster than you do.
Projects
AI Agent Minder
Open Source
A lightweight project governance layer for Claude Code — slash commands, markdown templates, and minimal hooks that add structured planning and decision tracking without getting in the way.
Claude Code Node.js Markdown
Airdrop Architect
commercial
SaaS platform for tracking crypto airdrop eligibility and DeFi points programs — Telegram bot plus web dashboard, built on Azure and C#/.NET.
C# .NET Azure Functions