@PostConstruct and @Produces are the same. They perform complementary functions. For a complex usecase, a combincation of @Produces with a Qualifier, Scope, and depending on whether you want data initialization or not -- @PostConstruct will be required. For ex, look at the following Memberlisting bean:

@SessionScoped
public class MemberListProducer implements Serializable {

     private static final long serialVersionUID = 1L;

     @PersistenceContext
     private EntityManager em;

     private List< Member > members;

     /**
      * @Produces creates a session scoped member list that can be injected in any bean 
      * @return
      */
     @Produces
     @Named
     @CurrentMembers
     public List< Member > getMembers() {
             return members;
     } 

     public void onMemberListChanged(@Observes( notifyObserver = Reception.IF_EXISTS ) final Member member ) {
          retrieveAllMembersOrderedByName();
     }

     @PostConstruct
     public void retrieveAllMembersOrderedByName() {
          CriteriaBuilder cb = em.getCriteriaBuilder();

          CriteriaQuery< Member > criteria = cb.createQuery( Member.class );
          Root< Member > member = criteria.from( Member.class );
          criteria.select( member ).orderBy( cb.asc( member.get( "firstName" ) ) );
          members = em.createQuery( criteria ).getResultList();
     }
}

The above SessionScoped bean maintains a list of members for a session. If the list needs to be accessed via a search screen then the data must be initialized, typically with @PostConstruct. It's job ends there. From this point, @Produces ensures that #{members} or @CurrentMembers retrieves the same list as long as the call is in the same session (unless you want a different scope). Seam 2's prior example of Factory resolves to something like this

@Produces
@ApplicationScoped
@Named
@CurrentMembers
public List< Member > getMembers() {
       return members;
} 

Hope that makes sense.

Typo in the first line above: @PostConstruct and @Produces are not the same