AI as Your System Design Copilot

An interviewer once shared an experience. A candidate was designing a chat application. Every answer was fast. Every diagram looked perfect. But something felt off.

When the interviewer asked:

“Why did you choose this database?”

The candidate paused.

Then said:

“That’s what the AI suggested.”

The interview ended shortly after.

Why?

Because AI is allowed as a tool, but thinking cannot be outsourced. This lesson is about using AI correctly, as a copilot, not as a replacement.

What AI Is Actually Good At in System Design

AI is very useful before and during design thinking, if used properly.

It helps you:

• Ask better requirement questions
• Discover edge cases you might miss
• Do rough capacity estimations faster
• Compare tradeoffs objectively

AI is good at expanding your thinking.

AI is bad at making final decisions for you.

That line matters.

Using AI for Requirement Questions

In interviews, the first step is always understanding the problem.

Beginners often don’t know what questions to ask.

This is where AI helps.

You can use AI to generate:

• Missing functional requirements
• Non-functional requirements
• Constraints you might forget

Example:
You ask AI:

“What requirements should I clarify when designing a chat app?”

AI gives you a list.

Your job is not to read it blindly.
Your job is to select what fits the problem.

In interviews, you say:

“I would clarify message delivery, offline support, scale, and latency.”

That sounds natural and confident.

Using AI to Find Edge Cases

Humans miss edge cases.
AI is very good at listing them.

For example:

• What happens if the same request is sent twice?
• What if a payment webhook arrives late?
• What if a user downloads a certificate 1 year later?

AI helps you think defensively.

But remember:
Just because AI mentions an edge case doesn’t mean you must solve all of them.

You prioritize.

That’s design.

Using AI for Estimations

Capacity estimation scares many people.

AI makes it easier.

You can ask AI to:

• Estimate QPS from DAU
• Estimate storage growth
• Calculate bandwidth roughly

But in interviews, never say “AI calculated this”.

Instead, you say:

“Assuming 1 million daily users, peak traffic could be around…”

AI helps you practice estimation so that you can explain it naturally.

Using AI to Compare Tradeoffs

AI is excellent at listing pros and cons.

For example:
SQL vs NoSQL
Sync vs async
Cache vs no cache

But AI often presents tradeoffs as equal.

Your job is to:

• Choose one
• Explain why
• Accept what you lose

Interviewers don’t want perfect answers.
They want clear decisions.

The Biggest Risk

AI sounds confident even when it is wrong. This is dangerous in system design.

AI may:

• Invent unrealistic limits
• Suggest tools that don’t fit the problem
• Overcomplicate simple systems

That’s why you need a verification habit.

A Simple AI Verification Checklist

Whenever AI gives you a design idea, quickly check:

• Does this solve the stated requirement?
• Is this needed at the current scale?
• Does this increase complexity unnecessarily?
• Can I explain this in simple words?

If you cannot explain it simply, don’t use it.

The Core Prompt Pattern

This is the most important part of this lesson. You will use this pattern every time you use AI for system design.

Assume

Tell AI your assumptions. User scale, region, budget, latency goals.

Ask

Ask AI for ideas, edge cases, or options.

Decide

You choose one approach.

Justify

You explain why you chose it. This keeps you in control, not the AI.

Example: Using the Pattern

• You want to design a certificate system.
• You assume moderate traffic and limited budget.
• You ask AI for storage and generation options.
• You decide to pre-generate certificates.
• You justify it by faster downloads and lower runtime load.

AI helps. You decide. That is the right balance.

Creating Reusable Design Prompts

Instead of random questions, you should use structured prompts.

A reusable design prompt looks like this (conceptually):

• Here is the problem
• Here are my assumptions
• List requirements
• Suggest architecture options
• Highlight tradeoffs
• Point out bottlenecks

You reuse the same structure for every system:
Chat app, payment system, AI tutor, monitoring tool.

This gives you:

• Consistent thinking
• Faster preparation
• Cleaner interview answers

Why This Matters in Interviews and Real Jobs

Interviewers can tell when someone is:

• Thinking independently
• Blindly copying AI output

Using AI correctly makes you stronger. Using AI blindly makes you unreliable. The goal is not to hide AI. The goal is to think like an engineer who uses tools wisely.

Key Takeaway

AI should:

• Expand your thinking.
• Speed up preparation.
• Improve coverage.

AI should not:

• Replace your decisions.
• Control your design.
• Speak for you in interviews.

You are the designer. AI is just your assistant.