The Corpus

When I started giving talks on AI in 2017, it was theory. Theory helps but doesn’t solve business problems — because the mess of organizational reality isn’t in books.

There are hundreds of thousands of people opining about AI who have never created with AI. Who have never grappled with infrastructure choices, fixed endless permissions snafus, or wrangled data less well organized than a landfill. In my case: 35 years of thinking strewn across thousands of folders.

I wanted to feel the same pain my clients feel. And almost weekly, I’ve felt the joy of building something real with my own hands. That isn’t in books or webinars.

Then I built this.

One person. No team. No engineering degree. No budget. I hadn’t touched a codebase in 35 years. The pipeline, the chunking, the embeddings, the query layer — built using Claude, from scratch, in a few weekends.

The engineering isn’t a footnote. It is the argument.

This isn’t a case study from someone else’s company. Not a prediction about what AI will enable. This is what happens when the gap between idea and execution collapses.

Ask about “human responses to organizational change” and any AI will give you the change curve — a concept thoroughly debunked by scholars in the field. Every search returns the consensus. The SEO-optimised. The algorithmically safe. AI trained on that body of work inherits all of it: the hedging, the mediocrity, the institutional caution dressed as objectivity.

Most AI is trained on the internet. This one was trained on books and research papers.

A body of scholarship with a 30-year evidence base, peer-reviewed foundations, and a documented intellectual lineage. Positions taken, defended, and revised in public, across eight books and dozens of white papers. Not the average of what everyone thinks about change, AI, and leadership — but what one person has argued, tested, and staked their reputation on.

You want an antidote to mediocrity. And one you can push back on.

To get to grips with an author, you buy their books, spend hours reading them, or use the index and hope your interests are covered. Across eight books and dozens of papers, that is an oppressive and expensive project.

Go to Amazon instead. Look up any business book. You get: “A groundbreaking exploration of leadership in the modern era, essential reading for executives navigating change.” Written by a publicist. Tells you nothing.

Here, you can ask: Does Paul actually have evidence for his claims about culture change? Or: How does Adaptive Adoption differ from Kotter — and why does it matter? Or: What is the link between information disorder and organizational culture? Or simply: Where is he wrong?

Real interrogation. Cited sources. No puffery. A library that talks back.

This is for researchers, serious readers, and anyone who wants to stress-test an idea rather than consume a summary.

This corpus contains work that has never been published.

The Great Collisions is a five-volume project on the deepest structural tensions of our time — the points where science, philosophy, economics, and technology are pulling in opposite directions and something has to give. Two volumes are queryable here, before they exist as books.

Including Oracle Science: what happens to science when AI predicts outcomes without generating explanatory theory? When the model is right but no one knows why? When the scientific method — hypothesis, mechanism, explanation — becomes optional?

A researcher can interrogate unpublished frameworks. A publisher can peer inside work in progress. A journalist can ask questions that haven’t been answered in print yet.

Some of this thinking doesn’t exist anywhere else on the internet. Query it here first.