Hermite®
Hermite® is an integrated platform for hit discovery and lead optimization powered by AI, physics and high-performance computing. With web-based molecular visualization and flexible cloud resources from AWS, it offers a ready-to-use and versatile platform for efficient and scalable computational drug discovery.

Leverage GPU acceleration, AI-driven algorithms, and parallel computing to deliver results 10–100× faster than traditional tools
Run simulations instantly via any browser—no installation or coding expertise required. Intuitive 3D visualization and drag-and-drop workflows streamline research for all skill levels.
Scale seamlessly from single-node experiments to enterprise-level workloads with AWS-powered infrastructure. Our smart scheduler dynamically optimizes compute costs without sacrificing speed.
Enterprise-grade security with end-to-end encryption, role-based access controls, and compliance-ready environments.

Lead Optimization
Free energy perturbation (FEP) predicts drug-target binding affinities at ±1 kcal/mol – a gold standard for lead optimization. Yet traditional FEP tools are often costly, complex, and inaccessible to non-experts. Hermite® Uni-FEP overcomes these barriers with an enterprise-grade platform that balances accuracy, speed, and ease of use for teams at all skill levels.
- High-Performance
Accurately predict binding affinity in as low as 4 hours.
Evaluate up to 1,000 compounds daily via parallel cloud computing.
- Robust
Validated across 100+ systems and 100,000+ perturbation pairs*.
>99% success rate for common use cases.
50% less manual intervention for complex systems.
- User-Friendly
End-to-end workflow with graphical interfaces tailor-made for FEP calculations.
Setup and run first task in 10 min for beginners; customizable for experts.
Intuitive and Integrated Workflow
Import from local files or built-in libraries
Automated protein & ligand preparation
Interactive 3D visualization & editing
Auto-generated perturbation graphs
Manual graph & atom-mapping adjustment
One-click submission with advanced parameters
Real-time progress tracking
Comprehensive analysis & error profiling
Exportable reports in multiple formats
Engineered For Precision & Scale
Customized GAFF2, supplemented by proprietary ML potential for an on-the-fly dihedral angle scanning and force field optimization workflow to tackle novel chemical structures. See our publications for details
REST2,Water Swap MC (buried water), Terminal Flip MC (multi-conformational ends), Conformation Exchange MC (macrocyclic)
.png)
Retrospective validation: 74% prediction within ±1 kcal/mol
compared with experiment.
Type of Tasks Supported
Use Cases




Flexible Pricing for Every Team
We are committed to democratizing FEP access without compromising performance. Choose the plan that suits your team’s need and goals.
Per perturbation pair pricing: Flat rate with volume discounts available.
No upfront commitment: Purchase credits, use as needed, and top up anytime.
Transparent billing: Only pay for completed calculations.
No limitation on bandwidth: Run 1,000+ tasks concurrently.
All-inclusive pricing: Save $$$$ versus maintaining in-house clusters and/or private cloud.
Flexible deployment: Use our SaaS platform or deploy on private cloud.
• Free training: Onboarding workshops and ad-hoc training sessions.
Virtual Screening Workflow
With modern chemical libraries rapidly expanding, virtual screening has become an essential tool for hit discovery. Uni-VSW offers a carefully-designed 12-step workflow, enabling researchers to efficiently prioritize and evaluate vast chemical space. Powered by a GPU-accelerated docking engine, Uni-VSW is capable of screening 100 million molecules in a single day.


ADME/T Property Prediction
Low-data scenarios and demanding model (re-)building processes have hindered the widespread use of ML in drug discovery, especially among smaller biotech companies. Uni-QSAR leverages the Uni-Mol pre-trained model to achieve high prediction accuracy and out-of-distribution prediction, even with minimal data. With automated parameter tuning, streamlined workflow, and built-in visualization tools, it presents a user-friendly solution for interpretable and efficient ADME/T property prediction.