[PRIMO Tech-a-Break] Using AI to Write Code Daily… Do You Know What Dev Level You’re At?

[PRIMO Tech-a-Break] Using AI to Write Code Daily… Do You Know What Dev Level You’re At?

Nowadays, AI has become an essential assistant in coding tasks and is increasingly integrated into many product development services — from code refactoring and test generation to analyzing complex logic. But the real question is: how are developers actually using AI?

At PRIMO Tech-a-Break, we’d like to introduce you to the 5 levels of AI usage — not just using AI, but using it effectively — so you can design and develop your product with maximum efficiency.

  But the real question we want to ask in PRIMO Tech-a-Break today is:
What level are developers really at when it comes to using AI?
Because there’s a big difference between just using AI and knowing how to use it well.

 

Level 1: Asking when you're stuck

You use AI like a faster search engine:

  • What’s wrong with this syntax?

  • How do I map an array more concisely?

  • How do I optimize this SQL query?

At this level, AI helps speed things up when you hit small roadblocks —
but it also comes with the risk of blind trust.

Level 2: Laying the groundwork, then building on it

This works best when you already know what to do but want to skip repetitive setup:

  • Generate a class or schema

  • Create basic test cases

  • Scaffold an API structure

If you’re doing full-stack work, you’ll likely use AI at this level a lot — especially for setting up projects quickly.

Level 3: Using AI as a sparring partner

Here, you’re not asking AI to write for you —
you’re using it to think with you:

  • Ask it to explain your logic back to you

  • Compare approach A vs. B

  • Spot holes in your reasoning

Personally, I use this level often.
It helps me see my own thoughts more clearly, just like when you explain your idea to someone else and suddenly understand it better yourself.

Level 4: Understanding intent, not just commands

Many think this only applies to chatbot developers —
but in reality, every dev needs this skill.

Example:

  • A user says, “I want the page to feel cleaner.”
    → They might not want you to remove everything — just make it feel less cluttered.

This skill is crucial in product development,
because what people say ≠ what they actually need.

(Just like I talked about in the last post on intent classification.)

Level 5: Let AI take over where it excels

Today’s tools are getting good enough that you can give AI full tasks:

  • Refactor code to use async/await

  • Fix failing tests and explain the changes

  • Rewrite logic to handle edge cases more robustly

If you set up your dev environment properly and know which tasks to delegate,
AI can take a lot off your plate —
but reviewing and making the final call is still on you.

So… what level should you aim for?

There’s no one-size-fits-all answer.
It depends on your work and your goals.

But here’s what I do believe:
Before you teach AI to understand others, you need to understand the “user” or the context first.

Because a developer’s job isn’t just to write correct code —
It’s to write code that serves real people.

 Thanks for reading this far.
If you’re using AI at other levels — or see things differently — feel free to share!
I believe how we use AI says a lot about how we think as developers.