Product · For research teams
Conjecture reads the literature, proposes testable hypotheses, designs the cheapest experiment that could break each one, and reports what failed.
Runs on the Lemma reasoning engine · Domain review before every project
Hypotheses
Experiments
Results
Literature first
Starts from what is already known, with citations for every claim.
Falsification plans
Proposes the experiment most likely to show a hypothesis is wrong.
Negative results
Keeps a record of what failed, so your team does not repeat it.
Shared workspace
Hypotheses, experiments and results on one board your whole team can see.
- 1ScopeWe review the research domain with your team before work starts.
- 2PilotConjecture works on one open question alongside your researchers.
- 3ReviewYou keep every hypothesis, plan and result, including the failures.
Question
Does the new catalyst work because of its surface area?
Illustrative example, not live model output.
engine
Today Lemma runs on leading foundation models, chosen for each task, with our own verification and evaluation layer on top. Our own models are in development.
- DecomposeBreaks a question into claims small enough to check one at a time.
- CheckVerifies each claim with code, a proof assistant, a search or a second model.
- ScoreGives every step a calibrated confidence, so people know where to look first.
- ReviewSends the steps it is least sure of to a person before anything counts.
- RecordKeeps the whole trace: every claim, check and decision, ready to audit.
Domains are approved one by one after a dual-use review. Ask us about yours.