The Tool

The Tool

What is it?

Quantisweb is a patented multi-objectives (targets) and multivariate Stochastic Approximation Optimization methodology and software which offers an alternative to traditional DOE methods; Expert Systems; Optimization, Simulation, Statistical software; Data Mining and last but not least “trial and error” to formulate or optimize a product recipe or a manufacturing process simultaneously.

  • Quantisweb determines the model (behavioural laws) of random systems with a minimal number of experiments.
  • Quantisweb generates an objective function (goal) using the desired product properties.
  • Quantisweb generates an optimal design (recipe) by optimizing the objective function and the model.


Value Proposition – How are we different?

  • Integrates Subject Matter Expert’s knowledge.
  • Considers desirable properties and their relative importance, as well as qualitative properties.
  • Reduces the time required to reach an optimal answer for a complex problem significantly by:
    • Generating a minimal dynamic DOE (mdDOE)
    • Handling critical, non-critical variables and associated constraints simultaneously.
    • Optimizing ingredients and processes simultaneously.
    • Using outputs of processes as inputs to subsequent processes.
    • Warning that product properties objectives might not be met without reviewing variable ranges and/or replacing ingredients.
  • Minimizes the data mining effort required when using historical data by:
    • Requiring the data from a minimal number of different production runs.
    • Validating the data for anomalies or corruption.
    • Estimating up to 15% of missing data for a property value.
  • Improve profits by reducing R&D and manufacturing costs, and time to market.


  • Starts with the end in mind. Leads user to determine desired product and process specifications at the outset.
  • Handles product and process variables and constraints simultaneously.
  • Eliminates Statistical Analysis. Chooses design of experiments method on its own from an internal library.
  • Deals with large problems. Generates np+1 experimental values for any number of variables regardless of the number of levels for each variable.
  • Provides X-Y patterns. Generates behavioral laws, using experimental values and associated results of experiments.
  • Chooses optimization method on its own from an internal library and optimizes combination of variables to reach.
  • Indicates existence of variability as well as parametric limitations.


  • The statistical error is negligible or under control.
  • The application does not want to understand the phenomenon.
  • The application does not create perfect behavioral law models – only patterns.


  • The way the application estimates the interaction of the X’s using a combination of parametric and/or nonparametric methods to generate Np+1 experimental values.
  • The way the application uses the Y’s measured from these experimental values to sequentially generate the behavioral laws based on the descending order of the importance of the Y’s.
  • The way the application generates a goal function that generalizes the loss function of Taguchi.

Classical Approaches vs. Quantisweb

classical approach versus Quantisweb