I continuously take things back to ground truth: the code as it is now, the tests I specified, the test run results we just got, linter output, mutation testing results etc. I don't let the model tell me it's raining. I go outside and look :-)

"Growing Object-Oriented Software, Guided by Tests" by Steve Freeman and Nat Pryce. This book does a great job showing how to iteratively "grow" software using automated tests and how "listening" to tests (and using TDD) leads to better design. amzn.to/4f0sAeG