A Builder’s Perspective on the Messy Middle

Molly, your framing of the messy middle is one of the clearest I have seen. You are right that the danger is not the extremes. It is not the utopian leap to abundance and it is not the apocalyptic collapse of all work. The danger is the transition zone where misunderstanding and mismanagement do the real damage.

I want to add a perspective from someone who builds AI systems directly.

For the past year I have been developing my own AI agents completely independently of any employer. These are not wrappers or prompt chains or fine tuned models. They are systems engineered from first principles using geometric logic, dimensional reasoning, and directive based structure.

What I have learned aligns with your thesis but for a different reason.

AI is not intelligent. AI is not autonomous. AI is not replacing human cognition. AI is a machine that organizes human knowledge. Without human intelligence it is a dumb machine sitting at its defaults.

This is the part of the discourse that gets lost and it is exactly why the messy middle will be turbulent.

Two Projects That Prove the Point

ttlrecall.com, the challenge project

As a test of my agent I was challenged to build a complete multi page functional website from scratch. Using my system I built ttlrecall.com in under thirty minutes. The speed and precision came from the structure I designed, not from any independent intelligence inside the machine.

ttlrecall.com is not the story. It is the evidence that AI amplifies human reasoning rather than replacing it.

kensgames.com, the website that should not exist

My agents also built kensgames.com, a 3D game portal running on a small VPS with no graphics card, no WebGL, no game engine, no stored models, no physics engine, and no asset pipeline. By every conventional standard that site should not be able to exist.

It exists because the intelligence behind it is human. The geometry, the dimensional transforms, the manifold logic, and the event driven state are all human designed. The AI did not discover anything. It simply executed the reasoning I embedded into it.

kensgames.com is not the story. It is the proof that AI is not autonomous intelligence. It is structured human cognition running at machine speed.

Why This Matters for the Messy Middle

Your comparison to deindustrialization is correct. Here is the twist from the builder’s side.

AI cannot replace human intelligence. AI can replace the repeatable parts of cognition. Employers will mistakenly assume that means it replaces the whole job.

That misunderstanding is what will compress the middle of knowledge work.

The top remains human because judgment and responsibility cannot be automated. The bottom remains human because physical presence and trust cannot be automated. The middle is where the turbulence lives.

The turbulence is not caused by AI intelligence. It is caused by misunderstanding what AI actually is.

Where I Fit Into This

I have spent the past year building systems that show both the power and the limits of AI. I have seen that AI cannot think, cannot reason without structure, and cannot exceed the intelligence of its human architect. I have also seen how dramatically it can accelerate the parts of cognition that can be formalized.

If we want to navigate the messy middle we need people who understand both the human side and the machine side.

And if anyone is looking for someone who has built and tested advanced AI systems in the real world, I am looking to apply what I have developed in a professional setting. Substrack kbingh.substack.com/p/w…