Hundreds of disease hypothesis generated computationally
Biology GPU
The biology resolution runtime for AI
We’ve built the experimental modality that enables AI models to reason in causal human biology
Pathways seen in action
in space and time
computable for AI
LLM
Multimodal Disease Models
Chat with Cells
Experimental Reasoning Loop
Experimental Reasoning Loop
BioGPU
Visual Biology Model Runtime
Multiple cell types across diseases, conditions and perturbations
Biology GPU
The biology resolution runtime for AI
Large-scale experimental infrastructure: Executes AI and omics-derived hypothesis directly in cells, across different disease-relevant cellular systems
Visual computing as a new biological modality: Observe pathway activity and target effects across millions of cellular images, interpreted by a model trained on 2B+ pathway images across ~140 cell types and 25 diseases
Parallel experimental execution: Resolve disease and target biology at scale, rather than one target at a time
BioGPU agent: Enables scientists or LLMs to design experiments, run visual compute, and iteratively test hypotheses against observed pathway activity
Visual Biology Trained Models
Large Visual Computing Model
Generated from 25 disease models trained in real projects with Pharma
Total: 2B+ process visualizations
Visual mRNA Biology Model
Domain specific model that visualizes mRNA biology regulatory mechanisms
BioGPU Agent
Enables LLMs to test biological hypothesis at scale inside cells
The missing experimental modality AI needs to resolve causal biology across pathways and processes in parallel
Under perturbation, cells activate many pathways at once.
BioGPU makes the full biological consequences of perturbation visible early enough to change decisions.
Visual Computing
inside cells
inside cells
Biological Processes and Pathways
Partner
AI Disease
Model
AI Disease
Model
Visual Biology
Causal AI
Disease Model
Causal AI
Disease Model
Genomics
Transcriptomics
Proteomics
Pathology
Cellular morphology
What AI can do with the Biology GPU
Disease Signature & Target ID
Discover pathway signatures - compare cellular systems
Discover novel targets that modulate pathway signatures
Target Validation
Understand a target’s biological function
Compare and rank targets by pathway modulation
Hit Identification
Discover compounds with pathway selectivity
Compare clusters for their biological activity - de-risk programs early
Hit To Lead
Elucidate compound’s MOAs
Optimize activity on core pathways
Lead Optimization
Optimize chemistry in a visual biology loop - optimize pathway activity
Predict off-target effects
Preclinical Development
Visual computing toxicity panels
Compare biology across human and animal models
+ Additional use cases in mRNA biology with a specialized mRNA neural network
Resolving disease and target biology
1
Hypothesis
2
BioGPU
See and compute on pathways inside real biology
3
Casual targets
Targets ranked by cellular activity, drivers separated from passengers