RXN uses AI trained on millions of chemical reaction records to suggest optimal synthesis routes, dramatically reducing manual effort, saving time and cost.
Upload molecules to determine reactions and ingredients
RXN is based on molecular transformer models that understand the natural language of chemistry, trained on over 3 million chemical reactions. Scientists can upload molecules to the cloud to determine reactions and ingredients. Results are accurate and fast, at a rate of ~7ms/reaction. Huge data sets containing millions of reactions can be mapped within a few hours.
Work with text-based or drawn representations of molecules and reactions as inputs
Convert chemical recipes to machine-readable instructions for use on autonomous lab hardware
Built-in model tuning, with domain-specific models ready in minutes
Third-party data sources made available to tune models on specific chemistry domains