Right now I’m building
Imperagen.
We combine AI, physics and automated experimentation
to build a recursive, self-learning engine for
reaction chemistry.
A MANIFESTO FOR LARGE REACTION MODELS
-
AI is crossing domains. LLMs learned language. LWMs are learning the physical world. The next frontier: the molecular world.
-
Chemistry remains stubbornly difficult. Progress still relies on exceptional people, long campaigns and a great deal of trial and error.
-
Closing the innermost loop. AI proposes. The lab tests in molecular reality. The results teach the model. The loop makes Large Reaction Models possible.
-
The unimaginable becomes inevitable. From new medicines and materials to cleaner processes and molecular repair—possibilities become reality.
-
A new interface to matter. Start from the outcome. Let intelligence find routes, test, learn and keep discovering.
Operator mode engaged.
This is my now.
At the molecular scale, proteins flex, atoms move, electrons shift and bonds form or break. These tiny events build the visible world.
A desired outcome could become the starting point.
The recursive self-learning path into the molecular world. AI proposes. Lab tests. Results teach the model. Model learns and decides what’s next.
The loop turns data into discovery.
Proteins are nature’s machines. Let’s teach them new jobs.
Chemistry will become programmable.
The map becomes a guide. Then an engine for discovery.
Better routes. Lower energy. Less waste. More impact.