Writing
Frameworks and field notes on running an AI-native marketing operation — for the people deciding where AI actually fits.
Frameworks & teardowns · The evidence floor
Too many affiliate links? I put the toplist on trial
Every affiliate operator has the same quiet fear: that a page full of 'Visit site' buttons reads as a sales brochure to Google and gets quietly buried. I spent a day interrogating that fear properly — Google's actual policy record, the conversion literature, 18 measured winning pages, and AI citation behaviour. The fear is aimed at the wrong thing.
9 July 2026 · 9 min read
Build-in-public · The Generate–Grade–Improve–Verify–Deploy loop
Most AI content tools are production lines, not intelligence systems
Marketing teams are running AI as a faster typing machine. The ones winning with it are running it as an intelligence system — a generate–grade–improve–verify–deploy loop with isolated agent stages. Here's the architectural difference.
27 June 2026 · 9 min read
Frameworks & teardowns · The three-layer AI search model
AI SEO strategy in 2026 is not what most people think it is
Most marketers are optimising for AI search the same way they optimised for Google in 2005 — chasing rankings, stuffing keywords, measuring positions. But AI search doesn't rank pages. It selects sources to cite. That's a different game entirely.
27 June 2026 · 9 min read
Frameworks & teardowns · The citability stack
Answer Engine Optimisation Is a Strategy, Not a Tactic
Most AEO guides hand you a checklist: add FAQ schema, write in plain language, use question-format H2s. That's not wrong — it's just not the point. Getting cited by AI is an authority problem dressed up as a formatting problem.
27 June 2026 · 9 min read
Frameworks & teardowns · The GEO stack
llms.txt is not robots.txt for AI. It's something more interesting.
The file is widely described as 'robots.txt for AI.' That framing is wrong in an instructive way — robots.txt is a restriction, llms.txt is a declaration. Here's what it actually does, whether it works, and what to put in it.
27 June 2026 · 7 min read
Frameworks & teardowns · Per-engine targeting
I asked five AI engines the same question. They cited different webs.
ChatGPT, Claude, Perplexity, Gemini and Google's AI Overviews don't cite the same sources for the same question. I measured how different — and the answer changes how you earn AI visibility.
24 June 2026 · 6 min read
Frameworks & teardowns · The source-type map
What AI actually cites — a 3,050-answer audit
I ran the same commercial questions through ChatGPT, Claude and Google AI Overviews and counted every source they cited. The engines don't agree with each other, and what they cite depends on your category. The data, and what it changes about AI-visibility strategy.
22 June 2026 · 8 min read
Build-in-public · The Compounding Memory Loop
The memory layer behind the AI operation I run
Almost every AI setup forgets everything the moment you close the tab — so you become its memory. Here's the architecture I built so mine remembers, and briefs me instead.
21 June 2026 · 6 min read
Transformation & operator economics · The architecture-over-headcount shift
Architecture over headcount: why AI-native leverage stops scaling with people
For codified work, AI drives the marginal cost toward zero. When that happens, output stops scaling with how many people you hire and starts scaling with how good your operating layer is. That's a different economics — and most marketing functions are still budgeting for the old one.
17 June 2026 · 8 min read
Governance & responsible adoption · Govern the blast radius, not the intelligence
AI governance that speeds you up — govern the blast radius, not the intelligence
Governance is sold as the brake on AI. Designed right, it's the accelerator — the thing that lets you safely say yes to autonomy. Here's the model, and why the 2025 data backs it.
14 June 2026 · 8 min read
Build-in-public · The AI-native maturity ladder
What an AI-native marketing operation actually looks like
Most teams bolt AI onto how they already work. A few rebuild around it. The gap between those two is about to decide who wins the next decade.
12 June 2026 · 8 min read
Frameworks & teardowns · The compounding test
Where Gen-AI actually moves marketing ROI — and where it's theatre
Most AI marketing spend produces a great demo and no measurable return. The difference isn't the model — it's whether the use case compounds. A test, and a map.
12 June 2026 · 7 min read