Colorectal cancer (CRC) is the third most common cancer worldwide. Although clinical outcome varies among patients diagnosed within the same TNM stage it is the cornerstone in treatment decisions as well as follow-up programmes. Tumor-infiltrating lymphocytes have added value when evaluating survival outcomes. The aim of this study was to develop a fully automated method for quantification of subsets of T lymphocytes in the invasive margin and central tumor in patients with CRC based on Deep Learning powered artificial intelligence. The study cohort consisted of 163 consecutive patients with a primary diagnosis of CRC followed by a surgical resection. Double-labeling immunohistochemical staining with cytokeratin in combination with CD3 or CD8, respectively, was performed on 1 representative slide from each patient. Visiopharm Quantitative Digital Pathology software was used to develop Application Protocol Packages for visualization of architectural details (background, normal epithelium, cancer epithelium, surrounding tissue), identification of central tumor and invasive margin as well as subsequent quantitative analysis of immune cells. Fully automated counts for CD3 and CD8 positive T cells were obtained in 93% and 92% of the cases, respectively. In the remaining cases, manual editing was required. In conclusion, the development of a fully automated method for counting CD3⁺and CD8⁺lymphocytes in a cohort of patients with CRC provided excellent results eliminating not only observer variability in lymphocyte counts but also in identifying the regions of interest for the quantitative analysis. Validation of the performance of the Application Protocol Packages including clinical correlation is needed.