Platform
One platform.Every step.
Target evidence, molecular design and experimental validation run as one system, so every result improves the next decision.
- Evidence engineRanks targets on genetics, omics, pathways and tractability — with every score explained.
- Generative designProposes molecules and sequences conditioned on the biology of each target.
- Closed lab loopAssay results retrain the models that chose each compound.
- Targets evaluated by the evidence enginePlaceholder
- 1,240
- Molecules generated and scored in silicoPlaceholder
- 3.2M
- Experimental success rate*
- 76%
* Share of AI-prioritised hypotheses confirmed in first-round wet-lab validation (owner-reported).
How it works
Five steps from hypothesis to candidate
Each step is a model and an experiment. Nothing advances on prediction alone.
- 01
Target identification
Multi-omics, human genetics and literature evidence rank gene–disease pairs by strength of evidence and tractability.
Evidence engine
- 02
Generative design
Structure- and sequence-aware models propose molecules against each prioritised target, scored on potency, selectivity and ADMET.
Design models
- 03
In silico validation
Docking, free-energy estimates and molecular dynamics filter designs before anything is synthesised.
Physics + ML
- 04
Wet-lab confirmation
Our laboratory tests the shortlist in orthogonal assays. Every result — positive or negative — returns to the models.
Closed loop
- 05
Candidate nomination
Lead series are optimised against a pre-agreed candidate profile before moving into IND-enabling work.
Decision gate
Capabilities
Three capabilities, one evidence trail
Explore how the platform ranks targets, reads omics data and connects evidence.
Rank the targets worth the next three years of your team
Our models weigh genetic association, expression evidence, literature momentum, tractability and safety signals into one explainable score for every gene–disease pair.
- Explainable scores: see exactly which evidence moved each target
- Re-weight by novelty, druggability or safety to match your strategy
- Every ranking traceable back to its source data
Design principles
Built to be questioned
Scientists do not trust black boxes, and neither do we. Every recommendation the platform makes arrives with the evidence behind it: the datasets, the model versions and the experiments that support or contradict it.
That traceability makes decisions faster, not slower. Teams can challenge a ranking, re-weight evidence to match their strategy and see the consequences immediately.
- Versioned data and models for every result
- Uncertainty estimates alongside every prediction
- Negative results retained and used for training
Put the platform to work on your disease area
We collaborate on focused target-discovery projects through to full co-development.