The Formulation Engine is a structured intelligence system built specifically for formulation-driven R&D to understand the scientific logic of your work from the ground up. Configure it using the real rules of your formulations
Mapping Your Formulation Logic
Before the engine can reason, it needs to understand your formulation space. Using the built-in configuration interface, you define your input variables - raw materials with percentage or absolute ranges, grouped ingredients with shared composition constraints, categorical choices like process conditions or additive types + your output targets, the properties you are trying to predict or optimize. Your input variables, output variables and constraints are imported from and connected directly to the Databases on the platform.
Refine & Create AI-Guided Experiments (AIGE)
AIGE takes the Formulation Engine beyond prediction and into active experimental design. By using the constraints and design from the mapping wizard, AIGE recommends the next most informative experiment to run based on what your data already shows. Alternatively, the DoE functionality can be used to smartly create trials from scratch. Fewer trials. Faster convergence. Smarter science.
AutoML & Model Evaluation
Once trials created by AIGE (or one that you bring) are performed experimentally, bring that data to the formulation engine and train predictive models automatically using AutoML for your properties. Every predictive model in the Formulation Engine is evaluated automatically without manual intervention. When a model is trained or retrained on new trial data, bodh scientific™ runs a full evaluation pipeline in the background and surfaces the results in your interface.
Each property gets its own model card. This gives R&D the context to decide when to trust a prediction, when to run more trials to strengthen a model, and when a property simply needs more experimental data before it can be reliably predicted. .
Property Prediction
When the models have been deployed, you can use the engine to predict target property outcomes. Use it to screen ingredient combinations, evaluate raw material substitutions, or assess the likely performance of a formulation before committing to an experimental trial run.
Formulation Optimization
Set your constraints and your desired outcome targets. The engine generates optimized formulation candidates that satisfy your specifications, with full transparency into the reasoning behind each recommendation. Built purposely for multi-parameter trade-offs where no single variable tells the full story.
