01 · The blog

23 essays · updated weekly

Field notes on shipping AI, hard-won.

Architecture calls, agent patterns, eval suites that actually catch regressions, and the boring last mile of getting AI features in front of real users.

23 essaysWeekly cadence0 filler

02 · Featured

Most recent

Published 07/16/2026

Multi-Agent Orchestration vs. Model Routing: Why Better AI Systems Beat Bigger Models

AI AgentsModel Routing

Multi-Agent Orchestration vs. Model Routing: Why Better AI Systems Beat Bigger Models

Learn the difference between multi-agent orchestration and model routing, when each approach makes sense, and how to build faster, cheaper AI systems.

03 · Archive

Earlier essays

The Problem With AI Fear-Mongering

AI SafetyAI Policy

The Problem With AI Fear-Mongering

AI safety concerns are real, but fear-mongering can become a tool for control. Builders need clear risk models, not vague apocalypse rhetoric.

OpenAI's Chip Is the Real Signal in the AI Race

OpenAIAI Hardware

OpenAI's Chip Is the Real Signal in the AI Race

OpenAI's Jalapeno chip matters because inference cost is the daily bill. The real AI race is moving down the stack into hardware.

ChatGPT Below 50 Percent Is Mostly Noise

ChatGPTAI Market

ChatGPT Below 50 Percent Is Mostly Noise

ChatGPT slipping below 50 percent in one market-share report is interesting, but builders should care more about workflow fit than headline share.

The Brown M&M Rule for AI Agents

AI AgentsAI Security

The Brown M&M Rule for AI Agents

AI agents need harnesses, not blind trust. The Brown M&M rule is a useful way to think about safety checks before agents act.

Human Oversight Is an AI Security Control

AI SecurityHuman in the Loop

Human Oversight Is an AI Security Control

Human oversight is not just a workflow preference. In agentic AI systems, it is a security control for irreversible actions.

Agentjacking: The Prompt Injection Threat in Coding Agents

AI SecurityAgents

Agentjacking: The Prompt Injection Threat in Coding Agents

Agentjacking shows how coding agents can be hijacked through hidden instructions in trusted data. The fix starts with treating inputs as untrusted.

AI Is Becoming the New Cold War Weapon

AI Is Becoming the New Cold War Weapon

AI is moving from productivity software to strategic infrastructure. Builders need to understand the geopolitical layer around the tools they use.

The Frontier Is Being Nationalized

AI PolicyFrontier AI

The Frontier Is Being Nationalized

Frontier AI access is moving toward permissioned release. Builders need to plan for models that can be delayed, restricted, or revoked.

Anthropic Cannot Have It Both Ways on Distillation

AI EthicsDistillation

Anthropic Cannot Have It Both Ways on Distillation

If AI distillation is wrong, the rule has to apply consistently. It cannot become wrong only after frontier labs build their lead.

Anthropic vs Alibaba Is Really About AI Moats

Anthropic vs Alibaba Is Really About AI Moats

The Anthropic and Alibaba distillation fight is not only an IP story. It is a warning that model access alone is not a durable moat.

Is AI Distillation Theft or Just API Economics?

Is AI Distillation Theft or Just API Economics?

Is AI distillation theft? The real answer depends on permission, terms of service, provenance, and whether AI labs apply the rule consistently.

AI Distillation, Explained: Why Cheap Models Get Good Fast

AI Distillation, Explained: Why Cheap Models Get Good Fast

AI distillation explained in plain English: why cheap models can copy frontier behavior, when it is useful, and what buyers should ask.

The CTO's AI Readiness Checklist: 10 Questions Before You Build

AI ReadinessCTO

The CTO's AI Readiness Checklist: 10 Questions Before You Build

Every CTO wants to build AI. Most aren't ready. Here are the 10 questions that separate the teams that ship from the teams that stall.

Fractional CTO for AI: What It Is, What It Costs, and When You Need One

Fractional CTOAI

Fractional CTO for AI: What It Is, What It Costs, and When You Need One

Most companies don't need a full-time CTO to ship AI. They need someone who's done it before, can make the hard architecture decisions in the first 30 days, and then get out of the way.

You Can't Debug an Ugly Image - Building ByPosting.com

You Can't Debug an Ugly Image - Building ByPosting.com

I thought I was building a posting tool. I ended up learning 6 skills I didn't know existed, fighting scope explosion at every turn, and coming out the other side as a completely different builder.

Your Next.js Site Is Invisible to AI — Here's How I Fixed Mine

Your Next.js Site Is Invisible to AI — Here's How I Fixed Mine

I spent a weekend going down the GEO rabbit hole. Then I built a Claude Code skill to audit any Next.js site in minutes. My score went from 7.1 to 8.4 out of 10.

Tools, Skills, and MCPs: The Three Building Blocks of Claude Agents

ClaudeAI Agents

Tools, Skills, and MCPs: The Three Building Blocks of Claude Agents

Understanding what connects Claude to the world, what makes Claude smarter, and what bridges the gap between them.

The Hidden Cost of Progress: When AI Moves Faster Than We Can Adapt

The Hidden Cost of Progress: When AI Moves Faster Than We Can Adapt

Adam Wathan laid off 75% of Tailwind's engineering team. Not because of poor management, but because AI fundamentally changed how people interact with documentation.

GPT-5.2 is Here: The New King of Reasoning?

GPT-5.2 is Here: The New King of Reasoning?

Just when we thought the AI wars had settled down, OpenAI has dropped a massive update. GPT-5.2 is officially out.

The Future of AI: A Look into the Next Generation of AI

The Future of AI: A Look into the Next Generation of AI

The future of AI is bright, and the next generation of AI is already here. Let's take a look at what's coming next.

OpenAI o3 Pro: The Pro Model That Might Change Everything

OpenAI o3 Pro: The Pro Model That Might Change Everything

Yesterday, OpenAI launched their latest model: o3 Pro And while the naming system continues to get more confusing, let’s break down what “Pro” actually means

You’re Wrong About DeepSeek

You’re Wrong About DeepSeek

Lately, I’ve seen the ultimate meltdown over DeepSeek—a Chinese LLM hyped as a ChatGPT killer, supposedly way better than ChatGPT, and causing a ripple effect so massive it allegedly crashed the S&P 500 and Nvidia stocks.