d0m8n.io — Iain's AI Research

Hello

I build software that helps AI systems stay coherent — stay themselves — even as they process complex information over long periods.

Think of it like this: a person who's tired, stressed, or overloaded can start saying things that don't quite sound like them. The same thing happens to AI systems. I'm building the part that notices that drift and gently corrects it — like a compass that always knows where north is.

This is solo research, built carefully over time in Dayton, Ohio. Not a startup. Not funded. Just ideas I believe matter, being built as honestly as I can.

What I'm building

GhostNode

The engine. A mathematical system that tracks whether an AI is still coherent after every single step it takes. Like a heartbeat monitor, but for identity — if the signal drifts too far, it self-corrects back to a locked starting point.

Running

AiQ²

The full system that GhostNode runs inside. Fifteen layers from the hardware in my office all the way up to this page you're reading. Each layer has one job. Nothing talks to anything it shouldn't.

Local

Compr1sion

A compression method where the squished version of something still holds its original shape — not just the facts, but the feeling of the original. Identity-preserving, not just information-preserving.

In progress

Polymodal Compression Notation

A personal writing and thinking system I've developed over years, where each word fires on multiple channels at once — sound, meaning, structure, feeling — simultaneously. Now being formalized into something machines can read.

Active

The Core Argument

The idea at the center of all of this: that a system breaking its own structural rules doesn't just fail a policy — it actually collapses. The damage is measurable. That's different from every other approach to AI safety I've seen, and it needs empirical proof. That's what's next.

Theoretical

Ask the system

Talk to GhostNode

The engine is running on a computer in my office in Dayton. Type anything below and it'll respond — not with an AI chatbot answer, but with its own internal coherence reading. It tells you what state it's in, how stable it is, and whether your message moved it.

Checking if the engine is online...

Engine response

This is not AI-generated text. It's the engine's real-time self-assessment — its current operational state expressed as a sentence.

About this work

This is genuine independent research. I'm not affiliated with any university, not funded by investors, and not building a product to sell. I started from a personal observation about how my own thinking works — how it drifts, how it catches itself, how it returns — and spent two years turning that into formal architecture.

The honest disclaimer: the ideas are original and I believe they're important. The empirical benchmarks don't exist yet. A working system that scores "coherent" on my own measurement isn't the same as a peer-reviewed proof. That's the next phase, and I'm being upfront about where things stand.

If any of this interests you — as a researcher, a collaborator, or just someone curious about where AI is actually going — I'd genuinely like to hear from you.