About us
Most AI gives you answers that are plausible. Science needs answers that are defensible.
We build agentic intelligence for science: every claim traced to evidence, uncertainty made explicit, and proprietary data kept under your control.
Why we exist
A new interface for science, with a dangerous habit
AI is becoming a new interface for science: a way to reason across data sources much faster than any human could. But it often hands you a confident, beautifully argued answer that’s quietly wrong.
On the Alchemic AI platform, every claim is traceable to the evidence behind it, uncertainty is treated as a signal rather than hidden, and because you run it over your own IP, nothing sensitive leaves your control without sign-off. We are building the trustworthy layer between powerful AI and the people doing science.
Our name
Keep the power. Build in the rigor.
Chemistry came from alchemy. AI for science is standing at the same crossroads today, and the history in our name is the argument, not decoration.
- 01
Alchemy
Humanity’s first attempt to transform matter. It gave us the laboratory, the apparatus, and the controlled procedure, mixed with magical thinking.
- 02
Chemistry
Modern science kept the rigor and discarded the wishful part, and made the dream of transforming matter scientific.
- 03
AI for science, today
Enormous promise, but too often confident answers without evidence, traceability, or a line between what is known, hypothesized, and merely conjectured.
Our mission
Build trusted agentic AI that orchestrates knowledge, computing, and experiments, removing the barriers between an idea and a discovery.
Our bet
The bottleneck isn’t just model capability. It’s epistemics.
Knowing what is true, why it is true, how confident we should be, and what evidence supports it. Science already has a discipline for this. We made it the architecture, not a disclaimer.
Every claim traces to evidence
Substantive claims link back to the evidence behind them, so experts can inspect and verify the reasoning.
Known, hypothesized, or guessed
What is established is kept apart from what is hypothesized and what is guessed, and the reasoning is reproducible.
Uncertainty is a signal
Uncertainty is quantified and surfaced as a first-class signal, never hidden behind a fluent answer.
Your data stays yours
You run the system over your own proprietary data. Nothing sensitive leaves your control without sign-off.
Five named agents
Reasoning the way science actually works
Specialized agents, each asking one of the fundamental questions of discovery, and each named for a figure from the history of alchemy and chemistry.
Geber
Discovery Strategist
“What should we do next?”
Van Helmont
Data Scientist
“What does the evidence mean?”
Hermes
Systems Engineer
“How do we execute?”
Maria
Protocol Developer
“What is the reproducible procedure?”
Philalethes
Evidence Verifier
“Is it true?”
“Lover of truth”, the hardest question.
Where we start
Sustainable chemistry, beginning with CO₂
Our first application is converting CO₂ into fuels and chemicals: one of the world’s most pressing challenges, and a proving ground that demands breakthroughs across catalysis, materials, processes, and manufacturing.
Solving it shows the platform can accelerate innovation where it matters most. Over time, the same trusted agentic infrastructure extends to adjacent areas.
- NowCO₂ conversion to fuels and chemicals
- NextSeparations
- NextBattery materials
- BeyondDiscovery across scales
North star
Enable the transition to a sustainable chemical industry by making discovery faster, more reliable, and more collaborative.
The new alchemy: transformation through science, at speed
If you want to bring AI to the domain sciences and industrial R&D, we want to hear from you.