Last week, I explained how S-comodels in the category of T-models allow computations running in an T-environment to be translated to computations running in an S-environment.
This week I explain a slightly different point of view: S-comodels in T-models allow computations with side effects from S to be translated into computations with side effects from T.
I illustrate this by exhibiting structures like deterministic automata, probabilistic automata, and hidden Markov models as S-comodels in T-models for suitable S and T, and compute the terminal such in some cases of interest.