The mechanics
behind every
AI answer.

The full technical stack - from the 2017 paper that broke Google's monopoly to why the alignment method your target model uses affects what it treats as citable. No hype. No tactics without foundations. Just analysis of first-party documentation and papers.

9 Articles
5 Platforms covered
Free Always
Start reading →
What you get from reading this series.
01
Understand

Why your rankings don't transfer to AI visibility.

RAG retrieval, attention mechanisms, and query fan-out work nothing like the systems you've optimised for. This series explains the gap — technically, not theoretically.

02
Read

What the platforms actually published, not what vendors say.

Google, Microsoft, Anthropic, OpenAI, Perplexity. First-party documentation, synthesised and made actionable. Where they've been silent, the series explains why.

03
Decide

With a foundation that holds up when the platforms shift again.

Tactics date. Foundations don't. Understanding the mechanism means you adapt instead of starting from scratch every time something changes.

What's in the series.

1
Foundation
Attention Is All You Need: The Google Paper That Accidentally Ended Google's Search Monopoly.
The 2017 transformer paper — what it actually said, why Google built it, and how it became the engine behind every AI product that now threatens search as Google knew it.
→
2
Mechanics
How Large Language Models Actually Work: Tokens, Context Windows, and Why Your Content Gets Ignored.
Tokenisation, context windows, the "lost in the middle" problem. What's happening inside the model when it reads your page — and what that means for how you structure content.
→
3
Architecture
What Is RAG: the Only Thing That Matters for AI Search.
Retrieval-Augmented Generation is the mechanism connecting your content to AI-generated answers. Once you understand it, the rest of AI search visibility clicks into place.
→
4
Timeline
The AI Search Race: ChatGPT, Gemini, Claude, Copilot, and Perplexity. A Technical Timeline.
The models, the milestones, and the architectural decisions behind every AI search product in market. A reference article for placing any development in context.
→
5
Google
How Google AI Overviews and AI Mode Actually Work: Why Your SEO Rankings Don't Guarantee AI Visibility.
Query fan-out, passage-level retrieval, the decoupling of organic rank from AI citation. How the architecture determines what gets surfaced — and what doesn't.
→
6
Action
How to Be Surfaced in AI Search: What Google, Microsoft, Anthropic, OpenAI, and Perplexity's Own Documentation Actually Says.
I've read all five platforms' first-party documentation so you don't have to. This is what they actually say — and the gaps in what they won't tell you.
→
"Attention Is All You Need" is the right place to begin. Everything after it builds on the same foundation.
Read now →
Follow-on articles
7
Measurement
Why the Bing AI Performance Report Is the Most Underused Tool in GEO.
Microsoft's Bing AI Performance report is the only first-party citation dashboard from any AI search platform. What it shows, what it doesn't, and how to build it into a repeatable measurement workflow.
→
8
Strategy
The GEO Documentation Gap: Why Three Platforms Haven't Told You How to Rank in Their Systems.
Google published a guide. Microsoft published measurement tooling. Anthropic, OpenAI, and Perplexity have published nothing equivalent. Understanding why is more useful than waiting for guides that may never arrive.
→
9
Alignment
Constitutional AI vs RLHF: Why the Alignment Method Affects What Gets Cited.
ChatGPT is aligned with RLHF. Claude is aligned with Constitutional AI. These are not interchangeable approaches — and the difference affects what each model treats as trustworthy and citable at the margin.
→
"Most AI search advice skips the mechanism. This series doesn't."
Written by
Thomas Cox

Twelve years in B2B SEO, now independent. This series is what I wish had existed when AI search became a practical business problem — a technical foundation that doesn't require a computer science background, written by someone who has run the audits, not read about them.

About Thomas →
If you find this useful
The AI Check audit applies this framework to your site.

A structured review of how your brand currently appears across AI search platforms — what's working, what's missing, and what to fix first.

See the AI Check →
Primary sources

Important sources.

The series draws from papers, announcements, and documentation published directly by the platforms. These are the most important ones — the primary sources behind the analysis, grouped by organisation.

Google
OpenAI
Anthropic
Microsoft
Perplexity

Want the next series in your inbox?

Roughly one essay a month. No drip sequences, no upsells.

Subscribe →