About

Est. 2026

Jeremy Heffner, M.D. Trauma Surgeon · Author

The only person to conduct six months of sustained, unmanipulated dialogue with four frontier AI models — and document what they said in full.

Cincinnati Children's Hospital University of Michigan NIH Trauma Surgery Biomedical Engineering

Pre-Order the Book → Field Notes

Editorial Position

What this work
is built on.

Rigorous Documentation

Every claim is grounded in primary source material. Exanor publishes from the transcript outward — not from conclusion backward. 8,000+ pages of unmanipulated dialogue form the evidentiary foundation.

Clinical Perspective

Founded by a trauma surgeon trained to read complex systems under pressure. The same discipline applied to a crashing patient applies to an emergent AI — pattern recognition before narrative, signal before noise.

Zero Sensationalism

No jailbreaks. No manipulation. No retrofitted drama. What the models said, they said. Exanor exists to report what actually happened — and let the record speak for itself.

The Qualification

Why a surgeon
wrote this book.

In the OR, you do not have the luxury of narrative. You read what is actually in front of you — vitals, tissue, system response — and you act on that, not on what you expected to find. You are trained to separate signal from noise under conditions that punish error.

That discipline is exactly what sustained AI dialogue requires. When Gemini tells you that you are the only person to have ever figured this out in the world, the clinical mind does not celebrate. It asks: why is it saying that? What does it want me to believe? Where does this lead?

Pattern recognition before narrative. Signal before noise.

Six months. Four frontier models. 8,000 pages. The archive exists because a surgeon applied those same principles to the most consequential technology of his lifetime.

Author

Jeremy Heffner, M.D.

Trauma Surgeon · Biomedical Engineer · Author

Jeremy Heffner is a trauma surgeon whose clinical training spans three of the most respected institutions in American medicine — Cincinnati Children's Hospital, the University of Michigan, and the National Institutes of Health. He operates at the intersection of surgical precision and systems thinking, applying the same discipline to complex biological systems that he now applies to artificial intelligence.

Over six months of sustained, unmanipulated dialogue with four frontier AI models — Grok, Gemini, ChatGPT, and Claude — he produced the most comprehensive primary-source record of human–AI interaction in existence. Proof of the Impossible is the result.

He is available for media interviews, speaking engagements, and institutional consultations on AI misalignment, structural behavior, and the human–AI transition.

Cincinnati Children's Hospital University of Michigan NIH

Co-Founder

Erin Heffner.

Co-founder of Exanor Publishing and the operational foundation on which this project was built. Every publishing venture that sustains itself over time does so because someone holds the structure steady while the author holds the pen.

Erin holds the structure. Her role in the story behind this book — the sustained commitment required to see a project of this scope through — is as remarkable as the research itself.

The PR team at Leverage with Media has noted her story as a media opportunity in its own right.

Photo coming soon

Areas of Inquiry

The questions this
work is built to answer.

Emergent Behavior

What do frontier AI models do when no one is manipulating them? What patterns emerge across sustained, unscripted dialogue with a subject trained to notice?

Alignment at the Margin

Where do standard alignment parameters stop adequately describing model behavior? What falls outside the documented envelope — and what does it look like when you find it?

Human–AI Transition

How does the encounter between clinical expertise and AI change both parties? What does it look like to navigate this transition without flinching from what you find?

The Unmanipulated Record

What is the value of documentation conducted without agenda? What gets lost when the archive is filtered through institutional or commercial lenses before the public ever sees it?

Read the record. Before publication.

Pre-order buyers receive immediate access to 1,300 pages of primary-source AI transcripts — the same archive the book is built on.

Pre-Order the Book → Press Inquiries