Typehints help communicate across time and space, to people who may never meet you or who might not be able to interrogate you about your code, so those people can understand how you expect the code to work.
Adding typehints is a succinct, more-complete form of communication than not-adding them. (It is rare, perhaps impossible, for all communication can be fully complete all the time.)
Further, you don’t know in advance which parts of the codebase are going to last for a long time, and which are going to be replaced in relatively short order. It’s probably better to to add the typehints when you know what they are, rather than to wait and see if you’ll “need” them later.
Typehints can be considered low-cost mistake-proofing against misunderstanding in an obvious place (i.e., where they are used), without having to look elsewhere (“just read the tests!” [groan]).