Driverless cars are within reach, while competent carebots are more remote. Yet we shall have all kinds of artificial agents in the foreseeable future. So we need to think about designing good artificial agents already today. I do not mean good relative to a specific purpose, or well designed so as to minimize harm to humans. Good artificial agents will be machines that are able to detect the possibility of harm or neglect of duty and customize a response to the situation. Such artificial agents will come to resemble moral agents like us. I shall briefly address three questions: Do we want artificial agents to make moral decisions? Do we need them to? And is that even possible, i.e. can it be engineered?