Your company may already know its next discovery.
Latent superintelligence: every experiment, every dead end, every decision your company has recorded, reasoning as one.
The discovery is
already in the building.
Not as a finished answer. As pieces, in different teams, years apart. Here are two of ours, and the thing that follows from putting them side by side.
Screened GRAS compounds against the enzyme’s cavity. Three candidates, all ordinary food compounds.
Read it ↗Project B · publishedA tearless onion nobody can prove still tastes rightSilencing the enzyme removes the tears. No sensory panel has ever compared the result to a normal onion.
Read it ↗Neither report says this. It is what is true when you hold both at once, and it is the kind of thing an organization already knows without knowing it.
Ask where you already work.
Answer from everything you know.
Can we stop onion tears without touching the onion?
Six stages,
one evidence trail.
What every investigation runs through, the onion one included. Real screens from Omic Discover. Open any of them full size.
Watch it
derive the answer.
The investigation above, recorded end to end. Every stage, every source, and the line where the evidence stops.
All research ↗- Established
Silencing the tear enzyme removes the tears. Settled science, and the reason the genetic route looked finished.
- Found by the first run
No published sensory panel has ever compared a silenced onion with a normal one. The flavor claim was an inference from chemistry.
- What followed
Leave the plant alone. Three food-safe compounds, cinnamaldehyde, eugenol and thymol, proposed to inhibit the tear enzyme while the flavor chemistry keeps running.
- Not yet done
Nobody has run the assay. The write-up includes the kitchen-scale experiment that would falsify it, and results will be published here either way.
The work behind it
Attest: a claim-native data layer for shared knowledge
Organizations share documents rather than facts, so every reader re-derives what a corpus says. Attest stores one row per source assertion, which makes corroboration a group-by and disagreement a first-class record. Measured on 150 biomedical concept pairs and a 50-paper corpus, with the costs reported.
Inspect the work ↗A survey of 3,974 archaeal biosynthetic gene clusters, and a substrate predictor for the fraction tools can see
Archaea are treated as an underexplored source of natural products. We surveyed 1,681 archaeal genomes, catalogued 3,974 clusters, built a substrate predictor for them, and tested the two usual explanations for the detection gap against matched bacterial controls. Neither survived.
Inspect the work ↗PRESTO: one model for binding affinity, and where it stops working
Affinity prediction, de novo binder ranking and mutational effects are treated as one problem by practitioners and as three by the tools. PRESTO is evaluated on all three, roughly halves the error of the established contact-based predictor, and reports the two tasks where it fails as findings rather than omissions.
Inspect the work ↗Bring us your
hardest question.
Nothing gets learned twice. Every answer goes back into the fabric, so the next question starts from everything the last one found.