Testing
Pengujian
The cheapest way to know a change is safe: a check that runs without a human reading anything.
Testing is where AI-assisted work pays for itself. A test is a written claim about behaviour that a machine checks, which is exactly the property that generated code tends to lack.
The order matters. Writing the test first gives the model a precise specification and gives you a way to know the change worked. Writing it after tends to produce a test that asserts what the code happens to do, which passes forever and catches nothing.
The remaining gap is that tests only cover the cases you thought of. An agent writing both the code and the tests will cover the cases it understood, and skip the ones it misunderstood. Keep some tests written by people, and keep integration tests that run the real thing.
Related terms
Code Review
A second pass by someone who did not write the code, looking for what the author missed.
Vibe Coding
Writing software by describing what you want and accepting the result without reading it closely.
Evaluation
Measuring whether the system does the job, on a fixed set of cases, repeatedly.
Regression
A change that made something worse. The reason to keep tests at all.
Agent
A model with tools, a system prompt, and a loop that lets it act and observe results.
Want this working inside your team?
It starts with one conversation about your workflow.