We are using this exact technique on the Amplication server where we automatically generate the models for the clients from the graphQL schema on the server. When you build your app with Amplicaiton you anyway get a generated code including the models both for the client and the server so you get exactly this OOB.