The role of AI in coding
AI coding is most helpful when used to speed up boring work and explain confusing things. You still set the direction: what is built, for whom, and how the result is tested.
A good request structure
A good request usually contains four things:
- Purpose: what you want to achieve.
- Constraints: the technology or rules to follow.
- Format: the desired form of the output.
- Done criteria: how you know the result is correct.
Example: building a page
Suppose you want a simple greeting card page. The request could read: "Create a single HTML page containing a title, one welcome paragraph, and one button. Use CSS in the same file, a soft green color, and no extra libraries."
# Follow-up request after the first result
# "Limit the content width to a maximum of 640px
# and give comfortable spacing on phone screens."
Fixing errors together with AI
When an error appears, copy the message exactly as it is and explain what you expected. An effective pattern:
# "I ran this file and got the message:
# 'X is not defined'. I expect the button
# to be clickable. Please explain the cause
# and how to fix it."
Security notes
Check every piece of code before using it, especially code that sends data over the internet. Never write credentials inside the code. For forms, remember that client-side validation can be bypassed, so checks must still exist on the server.
Ask via WhatsApp and we'll explain it in plain language.