Level-3 Fitness-For-Service (FFS) assessments for metal loss in pipelines, as defined in API 579, rely on finite element analysis (FEA), which is computationally intensive and time-consuming. This study presents a novel method to accelerate the FFS process by integrating a convolutional neural network (CNN) algorithm. The approach starts with collecting ultrasonic (UT) thickness data, which is then used in FEA to compute maximum allowable working pressures (MAWP) and generate pass/fail labels by comparing MAWP with the design or operating pressures. A feature set derived from the thickness data is used to train a CNN model to predict FFS outcomes, employing regression to estimate MAWP and classification to determine pass/fail results. Once trained, the model can perform near real-time assessments on new UT scan data, eliminating the need for repeated FEA simulations. This method significantly reduces the time and computational resources required for Level-3 FFS evaluations, enhancing both efficiency and cost-effectiveness.