When developing Legal AI playbooks – as discussed in the previous edition of the Legal Informatics Newsletter – we encounter some key design questions. One of the central issues is reliable output ...