Screening and identifying eligible patients for clinical trials relies on manual electronic health record (EHR) review. That can be time-consuming, labor intensive and prone to human error. This results in missed opportunities for clinical trial enrollment: fewer than 5% of adult cancer patients enroll in clinical trials. Automated screening for clinical trials could improve efficiency and recruitment to trials. We are testing this hypothesis in partnership with Mendel.ai, using natural language processing (NLP) to extract “unstructured” EHR data elements (e.g., free-form text fields from physician progress notes, pathology reports, etc.) key to determining cancer trial eligibility. We are testing an “AI-in-the-loop” intervention to increase identification of eligible patients, saving time for research staff and boosting trial enrollment, particularly among diverse care settings and patient populations. Novel solutions that rely on human-machine collaboration may help unlock the potential of machine learning in clinical trial operations.