Faster paths from evidence to the next experiment.
CITADEL Corporation builds AI-native research intelligence. Our mission is to shorten the distance between what the world already knows about a disease and the next experiment worth running.
Knowledge grows faster than anyone can read it.
Biomedical publishing, trial registration and molecular databases all keep growing, and each lives in its own system. The cost isn't only time. Promising leads sit unnoticed because nobody had the hours to connect a paper, a trial and a protein entry.
We think AI can do the reading and the cross-referencing, provided it shows its work. So we build tools that query real sources live, keep every claim linked to its record, check the AI's answer with rules it can't talk its way around, and stay open about what they can't do.
How we build.
Sources over assertions
Every candidate comes from retrieved records, and every brief says which sources answered. We'd rather show a gap than fill it with a guess.
Check the AI
Rules-based safety checks run after the model, so a withdrawn drug can't be presented as a top candidate.
Humans decide
Our output is a starting point for expert review. Researchers' ratings and notes shape how the system improves.
Honest limits
We publish our methods and limitations and run connector health checks in the open on our status page.