Primary care in the UK faces increasing workloads and limited resources, making it an ideal environment for the effectiveness of artificial intelligence (AI). The potential for AI to improve diagnostic accuracy through decision support systems or image analysis is clear. However, implementing these tools into primary care workflows is challenging. Major challenges include integration into the electronic health record (EHR), managing automation bias, and disrupting the doctor-patient relationship. The TRICORDER trial in primary care shows that these challenges need to be incorporated into study design using pragmatic trials. Implementation of an AI stethoscope in routine care did not significantly reduce heart failure detection (IRR 0.94 [95% CI 0.86-1.02]). However, the AI stethoscope independently increased the detection of heart failure, atrial fibrillation, and valvular heart disease. A pragmatic randomized trial design generates real-world data to overcome barriers to implementing innovations in healthcare.[3]