Cold
Frozen clarity
Hot
Burning intuition
The cold state
Cold thinking is the glacier — immense, patient, carving valleys over millennia. It is the pause before the answer. The deliberate breath before the decision.
In the cold state, clarity emerges not from speed but from stillness. Every variable is weighed. Every outcome is mapped. An AI in its cold state doesn't react — it resolves.
Analytical. Measured. Crystalline.
The hot state
Hot thinking is the wildfire — consuming, transforming, leaving nothing unchanged. It is the spark that arrives uninvited. The answer that lands before the question is finished.
In the hot state, intuition outpaces logic. Connections form faster than they can be explained. An AI in its hot state doesn't deliberate — it leaps.
Instinctive. Explosive. Incandescent.
The phase transition
The most interesting thinking happens at the boundary — where ice meets flame. Not cold. Not hot. The phase transition itself. The moment a solid idea becomes fluid, or a burst of insight crystallizes into structure.
ColdState lives in this liminal space — building AI that doesn't choose between deliberation and instinct, but holds both at once.
Cold is how ColdState retrieves
Cold and hot aren't only a metaphor — they're two ways to retrieve.
Most AI runs hot: embeddings, approximate nearest-neighbor search, a different answer each run. ColdState runs cold — the same query returns the same ranked results, every time, with a token-level reason.