Online variational inference

I’m looking for something like this where the posterior for one run becomes the prior for a subsequent run, but with VI https://docs.pymc.io/notebooks/updating_priors.html

is that sort of thing available? Is it just a matter of using the fitted variational approximation as a prior?