01Problem → discovery

Biological crop protection, built with AI-native discovery intelligence.

This is the real first Bioandex discovery: a frozen, auditable starting point for product development.

Scientific knowledgeDiscovery intelligenceExperimental learning

02Evidence

Knowledge becomes a connected system.

Bioandex connects curated observations, identities, literature and structure context—instead of simply searching documents.

36Pinned evidence records
12Target hypotheses
7Represented sources
curatedPHI-base
publicationLiterature
proteinM3AVQ5
structureStructure context
relationshipMechanism

Pinned evidence universe

A curated target-centered view. The full graph stays available under the hood.

03Discovery

12 biological targets identified.

A target is a part of the pathogen's machinery that may be worth disrupting. Each target is ranked by frozen evidence—not by an opaque answer.

Priority score is an unvalidated prioritization score, not a probability of biological efficacy. Unselected targets have not “failed.”

05Portfolio

Bioandex doesn't just pick the top five. It chooses what to learn next.

Each experiment should produce information—not just a yes or no result.

ExploitTest what looks strongest.
ExploreTest what could teach Bioandex the most.
BioandexBatch 0015 prospective selections
2 High score1 High uncertainty1 High information1 Comparison

06Freeze

We make the prediction before we see the answer.

Before external results arrive, Bioandex locks its evidence, priorities and evaluation rules. It is like handing in the exam before seeing the answer sheet.

BATCH_001PREDICTIONS
FROZEN
FROZENOUTCOME_PENDING
Predictions12
Selected targets5
PolicyFrozen
FeaturesFrozen
BaselinesFrozen
Evaluation planFrozen
Outcomes0
Prediction fingerprintf88719f1…7341
TodayPrediction frozen
FutureIndependent result arrives
ThenPrediction compared with reality

07Historical replay

Learn from the past—without calling it prospective.

Historical cases can be replayed by hiding a known outcome and reconstructing what evidence was available at the time.

RETROSPECTIVE
Historical caseAaSSK1Alternaria alternata · Citrus x microcarpa (related: calamondin)
Evidence cutoff2016-02-10
Historical evidenceReconstructed view

Historical reconstruction is useful for development and demonstration but is not equivalent to a prospective test.

08Learning loop

Public knowledge gets us started. Proprietary outcomes make us better.

Every external experiment can create Bioandex data that competitors cannot simply download.

BIOANDEX
PROPRIETARY OUTCOMES
1Predict
2Test externally
3Observe
4Learn
5Predict better

Model evolution

TodayTransparent heuristicCurrent
NextTargetRanker V1After outcomes
FutureMultimodal discovery modelAfter more outcomes
More predictionsMore external outcomesBetter proprietary datasetBetter modelsBetter experiment selection
“The discovery advantage compounds with every real prediction-to-outcome cycle.”
BIOANDEX

The intelligence is built.
The first prospective portfolio is frozen.
Now we turn external outcomes into better product discovery.

The engine is how we build the products. From public knowledge to proprietary biological intelligence.

12targets prioritized
5prospective selections
36evidence records
0outcomes seen before freeze