This post isn’t about the usual “how to prompt AI” advice.
It’s about lesser-known habits that can seriously upgrade the quality of code you get from LLMs.
Tip 1 — Say you’re not satisfied (immediately)
If the generated code isn’t what you expected, say it clearly and early.
Why this matters:
- LLMs adapt within a conversation.
- When you signal dissatisfaction, you push the model to:
- Reconsider assumptions
- Apply stricter reasoning
- Explore alternative approaches
Often the second attempt is more thoughtful than the first one.
Tip 2 — Ask twice (even when it’s good)
Even if the solution works perfectly, try this:
“Improve this code.”
You’d be surprised how often AI:
- Simplifies logic
- Improves naming
- Enhances performance
- Reduces edge-case risk
Iterative refinement with AI is like having a senior reviewer on demand.
Tip 3 — Give the big picture (and real context)
AI performs dramatically better when it understands your architecture and code structure.
Instead of only saying:
“Write a method that does X”
Try:
“This belongs to the service layer. We follow a layered architecture: Controller → Service → Repository. Where should this logic live?”
Now the AI can:
- Place responsibilities correctly
- Avoid leaking business logic into technical layers
- Produce code that fits your design
But here’s where things become powerful.
Instead of manually describing the system, you can use an AI Prompt Builder that automatically enriches your request with real project context extracted from your codebase.
For example, the CppDepend AI prompt builder can provide:
- Dependencies between components
- Layer violations
- Code metrics (complexity, size, coupling, etc.)
- Type relationships and call graphs
This context can be injected into the AI prompt so the model doesn’t just guess — it understands the structure of your system.
Now AI is no longer just a code generator.
It becomes an architecture-aware assistant working with real data from your project.
Tip 4 — Step away and refactor yourself
After generating multiple AI-driven methods, pause.
AI optimizes locally, not globally. You must:
- Reorganize classes
- Merge duplicated logic
- Extract shared abstractions
- Clean design inconsistencies
AI accelerates coding — but you are still the architect.
Tip 5 — Ask AI to criticize its own code
This is one of the most powerful and underrated tricks.
After the code is generated, ask:
- “What are the weaknesses of this code?”
- “Where could this fail in production?”
- “What edge cases are missing?”
Now the AI switches roles: from creator → reviewer
This often reveals:
- Hidden bugs
- Performance risks
- Missing validations
- Concurrency issues
- Security gaps
You essentially get a free design review.
Final Thought
AI doesn’t replace engineering judgment — it amplifies it.
The best results come when you use AI as:
Generator → Improver → Critic → Assistant,
while you stay the architect.
