A novel minimization technique was developed. The proposed method is based on a Gaussian random search in the parameter space and it can handle wide range of problems, including bi-exponential curve fitting, reasonable time. It uses only function values, does not require gradients.