PURPOSE: More than 20% of prescription errors in hospitals are due to an incomplete medication history. Medication reconciliation is a solution to decrease unintentional discrepancies between medications taken at home and hospital prescriptions. It is a normalised clinical activity but it is time consuming. Medication reconciliation usually uses three sources of information for an optimised medical synthesis, one of which is the patient. A conversational robot for patients could be a solution to assist. Numerous digital applications are designed for patients and need to be tested for usability, satisfaction, reliability and time saved. METHOD: We analysed Max, a conversational robot for patients scheduled for surgery in Toulouse University Hospital, using routinely collected health data in three successive steps. We examined willingness, compliance and patient satisfaction of usability with a Likert questionnaire and measured the time spent with Max and without. Finally, the reliability has been explored. RESULTS: The three successive observational steps were assessment of willingness and compliance (79 patients), time saved (61 patients) and reliability of the tool (68 patients). 71% agreed to use Max after a telephone call but only 73% of patients completed Max entirely. Max was well received and the overall satisfaction of usability was high for ease of use, readability, relevance and number of questions. Max saved a few minutes by optimised medical synthesis compared with a conventional telephone call. However, the reliability appeared to be lower than the human conventional telephone call. Randomised controlled trials are needed to confirm this feasibility study. CONCLUSION: Max was appreciated by patients and appeared to be suitable for assisting pharmacists in medication reconciliation. The tool established the list of treatments taken by the patient at home but reliability appeared to be lower than a conventional telephone call, recommending a 'double check' on the patient's arrival.