Before chemistry had a periodic table, there was alchemy. Practitioners observed that certain reactions worked, accumulated recipes for useful outcomes, and reasoned about transformations in terms of qualities like “hot,” “cold,” “wet,” and “dry.” They were highly skilled at producing results, but they didn’t have a science.
The periodic table didn’t just organize the known elements. It revealed the underlying structure of matter and turned alchemy into chemistry. The same observations, the same substances — but now with a theoretical framework that explained why things worked, predicted what would happen in new situations, and enabled systematic manipulation at the atomic level.
AI is where chemistry was before Mendeleev
Modern AI development has the same character as pre-Mendeleev alchemy:
- Practitioners have accumulated enormous expertise in what works
- Scaling laws predict that bigger models tend to be better
- Benchmark scores tell you whether a particular recipe produced a good result
But there’s no fundamental science. We can’t predict from first principles why one model architecture performs better than another on reasoning tasks. We don’t know why models hallucinate in some contexts and not others. We can’t explain why a fine-tuning run that should help sometimes hurts.
What a periodic table changes
The Holosynthics periodic table doesn’t solve all of these problems. But it provides what chemistry’s periodic table provided: a structural framework that makes systematic investigation possible.
With 56 elements across 8 types, we now have:
A common language — instead of “this model seems to reason better,” we can say “this model has 27% higher Relation element density.” Precise, reproducible, comparable.
Predictive power — element composition predicts performance on specific tasks. Models with high Entropy element density are more cautious; models with low Entropy elements are more prone to confident hallucination.
Targeted intervention — instead of “fine-tune on more reasoning examples,” we can identify which specific elements are weak and address them directly.
Holosynthics is a field, not a product
The periodic table of chemistry wasn’t just a tool — it was the beginning of a science. Thousands of researchers built on it. The elements, their properties, and their relationships became the foundation of modern chemistry, materials science, and pharmacology.
Holosynthics aims to be the same thing for digital matter: an open, reproducible framework that the AI research community can build on. Our papers are published on Zenodo with full experimental details. The research behind the framework is published openly.
The 56 elements are the beginning. The science is what grows from them.