This work presents a novel methodology for the automatic output-only modal identification of linear structures under ambient vibrations named Intelligent Automatic Operational Modal Analysis (i-AOMA). The proposed methodology leverages the Covariance-based Stochastic Subspace (SSI-cov) algorithm for output-only modal parameter identification and encompasses two key phases. In the first phase, the SSI-cov algorithm is executed using quasi-randomly samples of the control parameters. The corresponding stabilization diagrams are next processed via Kernel Density Estimation in order to prepare a training database for the i-AOMA's intelligent core. This is a machine learning technique (namely a Random Forest algorithm) that predicts the optimal combinations of the control parameters involved in the SSI-cov algorithm. The second phase of i-AOMA rests on the iterative generation of quasi-random control parameter samples. If a sample is classified as feasible by the intelligent core of i-AOMA, then the SSI-cov algorithm is applied, while a new sample is considered otherwise. Such iterative procedure stops once a statistical convergence criterion is fulfilled. Hence, ultimate modal estimates are carried out from the stabilization diagrams, and relevant statistics are computed to quantify the effects of the uncertainties attributable to the variability of the control parameters. The entire Python source code for i-AOMA is made freely available through public web repositories. Some applications of the proposed methodology are finally illustrated.