You would, for example, go to a school and use an entire classroom of children for one group, another full classroom for the second group, etc. Another way to think of it is that you're using naturally constituted groups.
Intact sampling is a type of non-probability sampling used to produce results that can be generalised only by making very strong assumptions about the sample(s). An intact group is an already-formed group (e.g., church groups, political organisations, or - as Hendrika mentions above - classrooms of students). No selection procedure is used, but the entire group is used to represent some larger population. The validity of results from this kind of sample is determined by the process by which the group was formed. For example, a classroom with 100 undergraduates in a required introductory course is more likely to represent all university students than is a classroom that has 25 psychology doctorate students.