A conversation can propose a change.
You describe an outcome in ordinary language. An AI tool proposes code. You inspect the proposal, decide what to apply and test it in your actual file. That can help a beginner move from a blank screen to a small working example.
The phrase “vibe coding” is used loosely. In Bhala’s lessons, the learner still owns the idea, rules, acceptance tests and explanation. A fluent answer or a game that runs once is not evidence that every rule is correct.
Start with a brief you can check.
“Build me a fun game” leaves too many decisions implicit. A better brief might say: “In one local HTML file, show a score and a native Collect button. Each accepted action adds one point. No network or accounts. Explain the changed lines. Test two actions and expect score two.”
That brief has context, an outcome, constraints, one requested change and an acceptance test. When the test fails, you have a specific observation to discuss.
What can go wrong?
Generated code can misunderstand a rule, reset only part of a round, refer to a missing element or introduce an unnecessary dependency. Ask for small changes, preserve a working version and compare expected with actual behaviour. Do not paste credentials or personal records into a prompt.
Do I need an AI account?
The supplied Bhala examples can be inspected and adapted without one. If you use an AI service, choose an approved, age-appropriate tool with the necessary permissions. Understanding an example and testing a change are useful tasks even without a live model.
Choose a small next step.
If you want to understand manual page structure and file paths, explore Web Development 102. If game rules and player feedback interest you, explore Game Development 106. You do not need to buy or repeat every foundation before making a start.
Find your starting point →