/notes

Field Notes

Long-form notes on AI, search, and the work. Some are useful.

Everything Already Exists. Find the Angle.

There is no untapped category left. The concept does not need to be new. The angle does. Here's the difference and how to find yours.

Read article

The Visual Workflow Tax: Why n8n Costs You Twice

Every node you add costs you twice. Once to build. Once to change it. Here's what vibe founders pay when they confuse a prototype tool with a production architecture.

Read article

Nobody Wants Your Survey. Here's What To Measure Instead.

Surveys measure politeness, not intent. The only validation that matters is behaviour under friction. Here's what to track instead.

Read article

Build From What You Are, Not What You Want To Be

Every founder has an edge they're ignoring. The knowledge, access, and context built over years that no vibe-coded clone can replicate. Here's how to find it and use it.

Read article

Coding Is Commoditised. Distribution Isn't.

Every vibe founder is optimising for the wrong scarce resource. Building is free now. Getting people to care still costs everything.

Read article

The Vibe Founder Trap: Why Builders Ship Everything Except What Works

Vibe coding removed the last excuse not to build. So everyone's building. Almost none of it matters. Here's the pattern.

Read article

LLMO Explained: The Long Game of Brand Absorption

The deep-dive on Large Language Model Optimization. How brands get absorbed into model training, why the vocabulary wins, and how to measure a surface with no dashboard.

Read article

GEO Explained: How to Get Cited by ChatGPT and Perplexity in 2026

The deep-dive on Generative Engine Optimization. How citations get awarded, what content earns them, and the exact moves that put your name inside a chat answer.

Read article

AEO Optimization: How to Win Google AI Overviews in 2026

The deep-dive on Answer Engine Optimization. Position zero, schema requirements, and the formatting moves Google's AI actually extracts.

Read article

SEO vs AEO vs GEO vs LLMO: Which Should You Optimize For in 2026?

A field guide to the four-letter acronyms eating your search traffic. Spoiler: you don't pick one.

Read article

Systems Thinking for Building With AI: The Model That Ships

Everyone has the same models now. The people who ship reliable products aren't writing better prompts. They're thinking in systems.

Read article

How to Decompose a Product Idea Into an AI System

The theory says architect the system, not the prompt. This is the method. Five steps to turn one idea into stages, decide what the model does, and define the handoffs.

Read article

Prompts Are Not the System: Why AI Demos Break in Production

Your demo works because you fed it the input that makes it work. Production feeds it the input you never thought of. The gap is everything you didn't build.

Read article

The Feedback Loop Is the Product: The AI Moat That Compounds

Everyone calls the same model. The one thing a competitor can't copy is a system that has been learning from its own output for months.

Read article

Where AI Systems Break Inside Teams: It's Ownership

The pipeline works in the demo. Then a team owns it, and it breaks. Not on the model. On ownership.

Read article

Systems Over Willpower: The Rule Behind Training and Code

The same rule runs a training block, a diet, and a product. Not more discipline. Less reliance on it.

Read article
No notes match. Try a topic chip above, or clear the search.