Core Concepts
AI Workflow Patterns
How we actually use AI agents to build production systems. Multi-agent coordination (Evaluator, Builder, Verifier) and the Plan → Implement → Verify loop.
When AI Fails
AI isn’t magic. We share what breaks, why it breaks, cascading errors, and how to build guardrails that catch AI hallucinations before production.
Human-AI Collaboration
The best results come from humans and AI working together. Learn when to rely on AI (boilerplate, tests) and when to use human judgment (architecture, security).
Real System as Proof
We’re building a production-grade multi-tenant SaaS platform as the vehicle for testing these workflows. The system is the example; the AI workflow is the story.
The New Series: Autonomous Dev Org
After 10 weeks of building, we pivoted from an idealistic cloud of agents to a pragmatic continuous loop. Follow the latest architectural shifts.Episode 1: The Orchestration Problem
Why one AI isn’t enough, and how we closed 35 tasks in 24 hours using a continuous loop and bounded contexts.
Series Overview
The full arc: what we’re building, where we started, and where the autonomous dev org series is headed.
Latest Deep Dives
Episode 3: Blast Radius Awareness
The impact graph architecture — Tree-sitter, KuzuDB, MCP — that gives agents proactive awareness before they write.
Episode 2: Memory That Survives the Session
GitHub Issues as structured agent memory. The bead format and an MCP search tool give the loop context without adding infrastructure.
Standards Can't Live in Your Head
How Claude Code hooks create an enforcement layer that never forgets your standards.
Stay Connected
Start Here
Choose your learning path based on your goals.
Subscribe
Get weekly updates delivered to your inbox.
About Me
Learn more about my background and approach.
New content published regularly (articles weekly, video shorts twice per week).