The autonomous discovery funnel, in the open engine
Traditional pipelines break across tools, teams, and weeks of waiting. The engine collapses the entire process into a single workflow — eleven stages from target identification to patient stratification, run from one conversation with your AI assistant.
Target Discovery
108K associations
Target Validation
5 evidence streams
Literature & Landscape
14K papers · 2.4K patents
Known Actives
2.4M ChEMBL · IC50/Ki
ADMET Profiling
31 endpoints · NovoExpert
Compliance
8 jurisdictions · inline
Lead Optimization
30+ scaffolds
Molecular Docking
AutoDock-GPU · PLIP
Clinical Outcomes
AUROC 0.756
MD Simulation
GROMACS · MM-GBSA
Patient Stratification
56 pharmacogenes
Target Discovery
108K associations
Target Validation
5 evidence streams
Literature & Landscape
14K papers · 2.4K patents
Known Actives
2.4M ChEMBL · IC50/Ki
ADMET Profiling
31 endpoints · NovoExpert
Compliance
8 jurisdictions · inline
Lead Optimization
30+ scaffolds
Molecular Docking
AutoDock-GPU · PLIP
Clinical Outcomes
AUROC 0.756
MD Simulation
GROMACS · MM-GBSA
Patient Stratification
56 pharmacogenes
The discovery funnel
Each stage has its own page with the science problem, proof, and a one-sentence prompt.
Target Discovery
Identify viable drug targets from 108,000 genomic associations — before committing lab resources.
Target Validation
Stress-test the target across five evidence streams — omics, trials, literature, ChEMBL, competitive landscape.
Literature & Landscape
Search the full research landscape in one query — 14,000 papers and 2,400 patents indexed.
Known Actives
Pull measured IC50/Ki potency from 2.4M ChEMBL bioactive compounds to seed optimization.
ADMET Profiling
Eliminate unsafe candidates early. 31 endpoints via the NovoExpert model family, called against any SMILES.
Compliance
Avoid regulatory dead-ends — 8 jurisdictions screened inline via the hosted FAVES API.
Lead Optimization
Generate optimized candidates automatically. Every variant enriched with ADMET and compliance.
Molecular Docking
Validate binding before you run experiments. AutoDock-GPU with strain-energy validation.
Clinical Outcomes
Predict Phase I clearance probability before committing to IND-enabling studies.
MD Simulation
Confirm stability with production-scale dynamics. GROMACS GPU with hydrogen-mass repartitioning.
Patient Stratification
Identify which populations will respond — 56 pharmacogenes, 135K resistance variants.
Start a discovery pipeline
Eleven stages. One conversation. Install the engine and run your first funnel.