In the study, 181 adults (88 women and 93 men) with insulin-treated type 2 diabetes and sub optimal control (glycated haemoglobin HbA1c ≥7·5% 58 mmol/mol to ≤11% 97 mmol/mol) were randomly assigned (1:1) to the intervention group, in which patients used the d-Nav device in combination with health-care professional support, or the control group, in which patients continued with a standard meter for glucose monitoring but otherwise received identical health-care professional contact. The reported metabolic outcomes are similar to those reported in an earlier single-arm service evaluation assessment of the same device in a UK setting5 and a broadly similar system in which dose advice was fed back to health-care providers.6 As expected in the context of a clinical trial, both groups had frequent contact with study team members, with seven face-to-face or telephone contacts over the period of 6 months. ...importantly, how comfortable would patients and clinical teams be in allowing an algorithm to manage insulin dose titration?