You'll need to start your test automation with a stable version of the application. Teach TestComplete that version and then when things change, you'll find out because places in the test will fail.
TestComplete can adjust some things as it goes, but you really don't want it to be able to automatically adjust to all changes because some failures in the test are going to be bugs and then you would never see them because the test always passes.
Build your test on a known-to-be-good version of the app. When you run it on a new version, then there will be some failures, and a person needs to review those and decide if the test needs to be adjusted (this would be from an intended update of the code) or the test is correct and a bug needs to be reported.