Demand for colonoscopy to diagnose colorectal cancer (CRC) is high, which can lead to late diagnosis and reduced survival. In addition, the ongoing coronavirus disease 2019 (COVID-19) pandemic has curtailed endoscopic procedures. Possible future alternatives include magnetic endoscopes. However, magnetic manipulation is complex, and clinical translation has proven difficult. In a new study that aimed to improve navigational performance, machine vision was used to enable intelligent and autonomous control by non-experts of a magnetic endoscope during colonoscopy. The system involved image-based autonomous navigation of the colonic lumen, but under the supervision of an operator, and consisted of a magnetic endoscope controlled by an external magnet attached to a robotic arm. It was tested in both a benchtop and in an in vivo porcine setting — in the latter, duration travelled in the colon was comparable to standard flexible endoscopy. The researchers hope that future models will enable higher levels of autonomy and ultimately provide a clinically viable alternative to conventional endoscopy.