Noetic: Looking for Partners

I’m building Noetic, and I’m looking for the first people to build it with.

One question

Today’s best models reason step by step, write code and pass professional exams. Ask one why it decided, and you get a fluent story. Ask it where that story stops being true, and it has nothing to say.

So here is the question I’ve spent this year on:

What would it take for an AI to know what it actually understands?

Not to sound confident. Not to explain itself well. To know, and to tell you honestly when it doesn’t.

The questions behind it

• Can a model tell the knowledge it understands from the knowledge it merely uses?
• Can it know where its own answer stops being valid, before it’s wrong?
• Can it notice that the world has changed, before you do?
• Can its decisions be checked by someone who doesn’t trust it?

I think the answer to all four can be yes. Noetic is my answer.

Where it stands

Built from scratch, tested with its pass criteria fixed in advance and its tests built blind, and early results are encouraging.

Who I’d like to talk to

• Investors who back deep tech early.
• Design partners in regulated domains, where “why” and “I don’t know” matter.
• Researchers and engineers who want AI that can be checked, not just trusted.

Reach me on X or by email.