Biosurfactants are produced through metabolism of microorganisms (bacteria, yeast and fungi) and many applications are attributed to them. Biomass concentration is an important variable in biosurfactant production process, because is related to substrate consumption and production rate. This variable is collected by sampling and determined by off-line analysis with significant time delay, in most of cases. The present work was carried out the development of models based on artificial intelligence to predict biomass concentration faster than usual analytical method (dry weight). The process was performed in batch-bioreactor using waste substrate. Feedforward neural networks were compared to determine the best model due to the database set acquired from the bioreactor plant. Software MATLAB 2016b was used to implement artificial neural network. The available input layers were agitation, aeration, absorbance, glucose concentration, dissolved oxygen concentration, surface tension and surface tension dissolved in 10 × and 100 ×. The network topology was determined by the combination of parameters such as number of neurons, training algorithm and activationfunctions. The results showed that the models are appropriate to predict the biomass concentration profile with good agreement of R2 (0.988), sum squared error (SSE) was 0.081 and mean squared error (MSE) was 0.0001.