DescriptionLabby
Open App

Granular chemistry data
In one place and one format

One platform to manage, visualize, combine and learn from all your chemistry data, with real-time insights powered by concrete and agentic AI.

Open Dashboard

Catalogue data

log results dynamically across domains, quickly and safely.

feedbackExperiment DesignChemical Resolution Worklist Generation Synthesis Execution Data Write-BackTriage (Level 1a)Full Level 1Level 2Data TransformationML FeedbackInput: ML prediction or human hypothesisOutput: Planned experiment record in ELNArtifacts: Experiment, Reactions, Chemical refsInput: Abstract reagent listOutput: Physical bottles with lot numbersArtifacts: Chemical → Physical mapping, inventory checkInput: ELN experiment recordOutput: Robot-readable worklist fileArtifacts: Worklist file, deck layout configInput: Worklist + physical reagentsOutput: Products in barcoded vialsArtifacts: Process Instances, Container (physical)Input: Robot execution logsOutput: Updated Process Data in ELNArtifacts: Process Data (actual conditions)Input: All synthesis productsOutput: Hit / Ambiguous / NegativeArtifacts: Test Instances, triage Sample DataInput: Hits from triageOutput: Complete identity profileArtifacts: Test Instances, Level 1 Sample DataInput: Characterised materialsOutput: Application performance dataArtifacts: Test Instances, Level 2 Sample DataInput: Raw instrument filesOutput: ML-ready feature vectorsArtifacts: Layer 1 → Layer 2 → Layer 3Input: Accumulated datasetOutput: New experiment proposalsArtifacts: Model predictions, design constraints

Put Graphs Here

add static blurb.

0.230.670.420.890.550.710.340.780, 0.31α: 40%β: 35%γ: 25%1, 0.88α: 50%β: 20%γ: 30%2, 0.45α: 30%β: 45%γ: 25%3, 0.62α: 55%β: 25%γ: 20%4, 0.79α: 35%β: 40%γ: 25%5, 0.53α: 45%β: 30%γ: 25%6, 0.67α: 25%β: 50%γ: 25%7, 0.41α: 60%β: 15%γ: 25%

Learn

Across Domains

Analyze

Modes of Failure

Predict

Rates and Results

Put Molecule model here

add static blurb.

AI powered insights, powered by broad domain data

classify results with automatic analysis

deployConditions Feature VecClassifierPhase LabelHistoryTrain SplitModel Fit MetricsReagent ratios (Si/Al, Na/Si, H₂O/Si)Temperature, timeSolvent, modulatorOSDA identityGel composition ratio vectorMorgan fingerprints from OSDA SMILESNormalised numeric featuresRandom Forest / XGBoostTrained on labelled synthesis routes>70% accuracy on framework typePure-phase ZSM-5Mixed phasesAmorphous / Failed synthesisELN synthesis recordsLabelled phase outcomesHistorical experiment archiveStratified split by phase labelCross-validation foldsHeld-out test setHyperparameter tuning (grid/Bayesian)Cross-validated trainingFeature importance ranking (SHAP)Accuracy, precision, recallConfusion matrix per phase classROC-AUC

Put Clustering Node 3d Graph Here

add static blurb.