Frontier AI for astronomy.
We are an independent research lab using frontier AI to accelerate astronomy.
Neptune was found on paper first.
In 1846, Uranus was not moving the way it was supposed to. Most astronomers saw noise. Urbain Le Verrier saw a signal.
Working at a desk in Paris, Le Verrier calculated where an unseen planet would have to be to pull Uranus off its predicted path. He sent the position to Johann Galle at the Berlin Observatory. On the night of 23 September 1846, Galle pointed his telescope at that part of the sky and found Neptune, close to where Le Verrier said it would be.
The planet had been hiding in measurements astronomers already had. It took someone looking at them more carefully to find it. The lab is named for that idea.
First the math, then the light.
Fast tools, careful claims.
We build research tools on frontier AI to help astronomy move faster. Speed is only useful if the results hold up, so we work by a short set of principles.
Rigor over hype
We use precise language and never claim more than the data supports.
Show our work
Code, methods and results are documented so that anyone can reproduce them.
Record the negatives
Rejected results and failed experiments are kept, not hidden. They are data too.
Credit the giants
Our work rests on public missions, archives and open-source software. We cite them every time.
Automate the tedious
Machines handle repetition so that people can spend their time on judgment.
A second set of eyes
No claim is announced until someone other than the person who made it has reviewed it independently.
Get in touch.
For research collaboration, questions or press, write to us.
[email protected]