How to predict a model with multiple factors?

My question is probably trivial but I cannot figure it out...
I'm doing a glm with multiple continuous and categorical predictors.
In R, the summary() gives the estimates of factors' levels as contrast to the intercept (right?). So, If I'm modelling, let's say, diet and sociality as categorical factors, and I want to predict with the model the y of a species that is herbivorous and solitary... which of the intercepts should I use???
In fact I have an intercept for herbivorous and one for solitary species. R can predict the y if I provide data as predictors, so I wonder If R use one of the two intercepts or does something different.

Thanks in advance.

PS. I could create a mixed categorical variable (es. social_carnivore, but the sample would reduce)