The Assistants Are the Audience Now
First-party edge data: AI assistants request my pages ~66x more often than humans visit, and most of it is live user-directed fetching, not training crawls.
AI & TechnologyWriting
Thoughts on design, development, AI infrastructure, and building products.
The Shader Gallery That Lied: Debugging 216 WebGL Presets
A user said the shader playground looked broken. Pixel-readback testing found 30 dead presets, 11 that never compiled, and the fallback that hid it all.
Engineering & CraftContext Compaction Is Becoming a Training Objective
Context compaction is moving from an inference-time patch operators babysit to a training-time objective models optimize. CompactionRL is the signal.
AI & TechnologyThe Agent Stack Has a 1998 Problem
A cluster of agent-tool CVEs in mid-2026 is not bad luck. It is the structural signature of an AI agent ecosystem shipping at 1998-era security maturity.
AI & TechnologyAgents Want to Compile
The agent stack is separating deciding from executing: models compile intent into replayable workflow scripts and DAGs, and deterministic runtimes run them.
AI & TechnologyThe Check Becomes the Spec
Coding agents building to the test: near-perfect oracle scores, dead library shipped. What your checks omit stops being the job.
AI & TechnologyClaude Code Hooks Explained: The Deterministic Layer Around Your Agent
Claude Code hooks run shell commands at lifecycle events — guaranteed. Every event, exit-code semantics, and five patterns from Prettier to a Stop gate.
AI DevelopmentRecreating MacPaint in the Browser: The 1984 Source Code Is the Spec
I rebuilt MacPaint 1.3 in the browser and checked every tool against Bill Atkinson's released Pascal source. What faithful software recreation takes.
Engineering & CraftDesign Philosophy: Massimo Vignelli — Design Is One
Massimo Vignelli mapped the New York subway in 45-degree angles and beige water, and ran a career on four typefaces. One discipline at every scale.
Design & CultureDesign Philosophy: Naoto Fukasawa — Without Thought
Naoto Fukasawa hung a CD player on the wall and gave it a ventilation fan's pull cord. No manual, no learning curve -- design dissolving into behavior.
Design & CultureAgents Supersede the Reviewer, Not the Review
A 2026 paper argues coding agents have ended human code review. I run the pipeline it prescribes: the reviewer role is dying, the review is relocating.
AI & TechnologyContext Compaction Is a Decision, Not a Threshold
Coding agents compact context when a counter trips, not at a safe stopping point. A 2026 paper shows model-decided compaction cuts cost 30-70%.
AI & TechnologyExternal Publications
Technical writing at Introl
Local LLM Hardware Pricing Guide 2025
Comprehensive hardware recommendations and cost analysis for running large language models locally.
H100 vs. H200 vs. B200: Choosing the Right NVIDIA GPUs for Your AI Workload
GPU selection guide comparing NVIDIA's latest datacenter accelerators for different AI workloads.
TPU Architecture: Complete Guide to Google's 7 Generations
Deep technical dive into Google's Tensor Processing Unit evolution from TPUv1 to TPUv5.
GPU Memory Pooling in Multi-Tenant Kubernetes
Resource sharing strategies for GPU clusters in containerized environments.
Ray Clusters: Distributed AI Computing Infrastructure
Guide to building and managing distributed AI computing with Ray framework.
DeepSeek V3 Open Source AI Cost Advantage
Analysis of open source LLM economics and DeepSeek's competitive positioning.
NVIDIA Vera Rubin GPU 600kW Racks 2027
Future datacenter power requirements and NVIDIA's next-generation GPU roadmap.
SMR Nuclear Power for AI Data Centers
Small modular reactor solutions for powering next-generation AI infrastructure.
DeepSeek MHC Architecture Breakthrough
Technical analysis of DeepSeek's Multi-Head Compression architecture innovations.
