A single cell runs thousands of simultaneous reactions, builds proteins from blueprints, repairs its own DNA, maintains structural integrity under stress, and reproduces.
It does all of this without a general-purpose reasoning engine.
No prediction. No pattern matching. No generation. Every reaction works through shape — a molecule fits a substrate, the reaction fires. Thermodynamics and probability do the rest.
ATP synthase is a literal rotary motor, ten nanometres wide. A shaft spins inside a ring. Each rotation produces three ATP molecules. Your cells run about ten to the twenty-one of these per second. No one is coordinating them.
Kinesin is a protein that walks. Two legs, step over step, along microtubule highways, carrying cargo vesicles. Eight-nanometre steps. It looks like a person carrying a sack.
DNA polymerase copies three billion base pairs with roughly one error per billion. Then proofreading enzymes walk behind it and fix most of those.
None of these have a controller. No CPU. No orchestrator. They work because of shape — molecular geometry and thermodynamics. The right shape fits the right substrate, the reaction happens. Billions of these simultaneously, no coordination layer.
I build AI software. I use large language models every day. The models are the most capable, most expensive, most complex component in my system.
And the most complex system I know of — the cell — has nothing like them.
That gap is worth sitting with.