Hey,
Have you got a diagram of your setup.
What existing test coverage do you have?
There will be contracts between integration points between two applications. I would generally start with the high value endpoints, one that are subject to change frequently, or those which is they did change, would cause significant detriment.
Each of the applications should have a distinct unit test suite, where I would assume developers are performing component integration with mock representations of their dependencies.
It is these mocks that are subject to drift from the real implementation, and that is what you will be using contract testing for, to provide that feedback prior to implementing and deploying both sides.