Grommet insertion is a common otological procedure, where postoperative review remains essential to confirm grommet patency, assess for extrusion, and identify complications. These routine assessments place considerable demand on ENT outpatient services and create significant access challenges for families, particularly those in remote rural regions. Artificial intelligence (AI)–enabled assessment may support automated interpretation of otoscopic images to evaluate grommet position and patency, enabling local follow-up whilst also flagging high-priority cases for specialist review. Methods: Postoperative otoscopic images retrospectively collected from paediatric ENT services in the Northern Territory and Queensland were annotated for grommet presence, position, and lumen patency. An ensemble otoscopic image classifier was developed using convolutional neural networks and trained using 450 images collected from tele-health services at the Royal Darwin Hospital and the Deadly Ears program in Queensland. The model was evaluated on an independent test set of 212 images collected during grommet reviews performed at Logan Hospital between 2021 to 2025. Performance was assessed using 95% bootstrap confidence intervals (CI) to classify grommet positions, and determine lumen patency based on tympanometry. Results: 260 children were reviewed in postoperative grommet check clinics. Of these, a total 212 otoscopic images were collected. The model achieved an accuracy of 74.3% (95% CI 65.7–81.9%) for determining a grommet's position as in-situ or extruded. For grommets confirmed to be in-situ, it achieved an accuracy of 84.4% (95% CI 76.6–92.2%) for distinguishing grommets with patent lumen. Conclusion: AI-assisted analysis of postoperative otoscopic images shows promise as a supportive tool for grommet follow-up. When integrated into community-based pathways, such tools may enhance access to postoperative care, reduce unnecessary clinic visits, and alleviate pressure on ENT services in rural and underserved settings.