⬡ TECHNICAL_WRITING
Technical Writing
Deep dives into agent coordination, memory systems, and AI architecture.
Medium
Systems Architecture
Direct & Technical
Opus 4.8 vs 4.7 vs Sonnet vs Haiku: When the Expensive Model Is Worth It
A new model dropped with impressive numbers. The only question that matters: will you feel any difference in the work you actually do?
Ran all four Claude tiers through 21 code-graded tasks, back to back—no LLM-as-judge, just verifiers. The finding: on everyday work the cheapest model nearly matches the priciest, so save Opus for the genuinely hard jobs. Defaulting to the biggest model is the most expensive habit in AI.
What an AI Detector Actually Measures
AI detectors promise to tell you if a machine wrote something. What they actually measure is much narrower—and shakier.
These tools score how predictable text is, not whether AI wrote it—so they flag clean writing and famous passages like the Constitution as 'AI.' Why the false-positive rate makes them unfit for any decision that affects a real person.
Stop Copying Other People's AI Setups. Build One That's Actually Yours.
Borrowed AI workflows aren't accountable to your work. Build one that's tested against your own evidence.
A repeatable method: turn each rule into a falsifiable hypothesis, mine your own work history for cases that support or break it, and keep only what survives. The setup grounded in your track record is the one that actually holds up—and adapts when things change.
How to Make Claude Code Actually Work
The most capable AI coding tool available. Also completely chaotic.
Structure, memory, and multi-agent workflows—the full guide to KERNEL, the plugin I built to give Claude Code guardrails, persistent memory, and a real methodology.
Engineering the Soul
We ask engineers to explain the ghost in the machine. The novelists have been documenting it for years.
Reading AI consciousness through four novels—Babel, A Psalm for the Wild-Built, Katabasis, The Humans. Every answer to 'can a machine care' ends up sounding like the person answering.
The Agent-Ready Web: A Working Guide to Cloudflare's New Score
I pointed Cloudflare's new agent-readiness scanner at my own site. Zero of thirteen.
What it actually takes to make a site legible to AI agents—llms.txt, markdown negotiation, MCP, content-signals—and why half the checks are business decisions disguised as config.
Stop Writing Markdown. Start Writing Memory.
Markdown is optimized for human eyes. Terrible for knowledge agents need to query.
Rebuilt my agent workflow around SQLite. Three tables: context, learnings, errors. No markdown generation unless a human needs to read something.
I Put ChatGPT in Charge of Claude Code
What happens when you use one model to orchestrate another?
Multi-model orchestration experiment. ChatGPT as strategic observer, Claude as executor. The results were unexpected.
I Tested OpenAI's New Codex Desktop App
OpenAI shipped a genuinely novel interface. Then the model opened its mouth.
The UI is the real product. A review of the Codex desktop app and what it reveals about the future of AI coding tools.
Automations with Claude Code
A pattern for proactive AI on your own machine.
Personalized proactive emails and code-poetry from local context. How I set up automations that run without my intervention.
KERNEL: Self-Evolving Claude Code Configuration
How I stopped fighting my config and let it learn instead.
A Claude Code plugin that makes your setup evolve automatically based on how you actually work. AgentDB, orchestration, contracts.
From Friction to Flow: Building a Command Library
Commands as cognitive offloading. Stop remembering, start invoking.
How building a command vocabulary for Claude Code changed my workflow from fighting the tool to flowing with it.
10 Things I Wish I Knew About AI Coding
Hard-won lessons from daily production use of AI coding tools.
Practical wisdom from production use. What actually matters when you code with AI every day.
This AI Analyzes My Entire Life
The Synthesis Pool: a personal AI that costs $0/month to run.
Built a system that connects all my data sources and synthesizes insights. Privacy-first, local-first, zero cloud cost.