Your coral records stay where they are.
Source colonies, genotypes, nursery material, outplants and monitoring remain in the organisation's own system of record. CoralQuery Science reads them and never changes them.
View all Alchemy Loop products← Alchemy Loop portfolio
Governed answers for coral restoration science
CoralQuery Science
Ask the science, not the database.
CoralQuery Science gives restoration scientists one place to ask about genotypes, lineage, experiments, nurseries, outplants and monitoring—with the evidence behind every answer.
WORKING PROTOTYPE
ONE GENOTYPE · CONNECTED EVIDENCE
FIELDSource colonyLocation · lineage
LABGenotype RR-A017Batch · protocol · trial
NURSERYLiving materialStock · fragments
OUTPLANTMonitoringSurvival · bleaching
ANSWEREvidence, logic and caveats attached
The problem
Restoration science is spread across systems.
Genotyping results sit in lab files, nursery stock and outplants sit in operational databases, and experiments sit in notebooks and spreadsheets.
“Which genotypes survive well in the field but have never been heat-stress tested?”
Answering that question often means exporting, joining and cleaning data by hand—every time.
What it does
Three controlled parts, joined by the genotype ID.
A coral-first lab notebook.
Experiments, versioned protocols, tissue samples, genotyping batches and heat-stress trials carry independent sign-off, version history and an audit log. Genotype IDs are checked as they are entered.
Plain-language questions with evidence.
Answers can combine lab and field context, showing which system each part came from, the query logic and the caveats that affect interpretation.
Built for scientific integrity
Useful answers without erasing uncertainty.
The prototype is designed to make limitations visible before a number is treated as a conclusion.
Read-only by construction
Coral records are read, never written. The lab notebook is the controlled place scientists write, with sign-off and version history.
Evidence attached
Every answer shows its sources, query logic and caveats. Exported datasets retain their filters and sources.
Ambiguity surfaced
When a scientific definition has more than one reasonable interpretation, CoralQuery Science asks rather than guesses.
No flattering numbers
Early deaths remain in survival figures, unmonitored corals are reported separately and small samples are flagged rather than ranked.
Association is not causation
Correlations are labelled as descriptive, never presented as findings.
Human authority
Scientists interpret and decide. Notebook entries require independent sign-off from someone other than the author or last editor.
Questions scientists can ask
Follow a genotype across the work.
CoralQuery Science is intended for questions that need evidence from more than one stage of restoration science.
- 01
What do we know about genotype RR-A017, and where is living material today?
- 02
Which genotypes have never been heat-stress tested?
- 03
Which genotypes have fewer than five living fragments across all nurseries?
- 04
Show one-year survival by genotype for the 2025 outplant cohort.
- 05
Which outplants bleached in the 2025 thermal event and later recovered?
Five-minute product demonstration
Watch CoralQuery Science connect lab and field evidence.
The upcoming walkthrough follows a fictional genotyping batch from lab results to registered genotypes, traces one genotype from source colony to outplant with its heat-stress results, and shows plain-language questions answered with evidence.
Demonstration data only; no real organisation's records are shown. The narrated demonstration uses an AI version of Tom's voice.
Where it stands
A working prototype on fictional data.
It demonstrates a genotyping batch, follows a genotype from source colony to outplant, and answers a fixed set of questions that join lab and field records.
Fixed question types, evidence-backed answers and fictional demonstration data.
The LLMQuery engine, scientist-approved rules and evaluation against a restoration team's own data.
See it on your own questions
Bring the questions your team answers by hand today.
We will walk through the prototype and map those questions against your data.