Artificial intelligence is no longer a futuristic concept in software development—it’s already reshaping how engineers write, debug, and ship code. Over the past year, I’ve been working extensively with tools like GitHub Copilot, Claude Code, and Gemini CLI, and the shift has been dramatic.

The conversation is no longer about whether AI will change programming. It’s about how engineers must evolve to stay relevant, productive, and secure in this new era.

This article combines insights from the Mr. Will podcast, and—most importantly—my own hands‑on experience using AI coding assistants daily. If you’re a software engineer, educator, or tech leader wondering how AI will impact your workflow, this deep dive is for you.

Why AI Is Changing the Role of Software Engineers

AI coding tools can now:

• Generate large chunks of code
• Suggest architecture patterns
• Debug issues
• Analyze dependencies
• Perform security checks
• Write documentation
• Refactor legacy code

This means the traditional idea of “engineers = people who write code” is fading. Instead, the engineer’s role is shifting toward:

• Problem framing
• System design
• Prompt engineering
• Risk assessment
• Security review
• AI orchestration

In other words, the value of an engineer is moving up the stack—from typing code to directing intelligent systems.

My Real Experience Using AI Coding Tools Daily

I’ve spent the past year integrating multiple AI tools into my workflow:

• GitHub Copilot for everyday coding
• Claude Code for deep reasoning, refactoring, and understanding complex logic
• Gemini CLI / Coding Partner for security checks and dependency analysis

Each tool has strengths and weaknesses, and using them together has given me a clearer picture of what AI can (and cannot) do.

✅ GitHub Copilot: The Reliable Daily Coding Partner

Copilot is excellent for:

• Boilerplate
• Repetitive patterns
• Quick function drafts
• Inline suggestions
• Speeding up everyday tasks

It doesn’t overthink. It just writes. And that’s exactly why it’s so useful.

✅ Claude Code: Brilliant… Until It Gets Stuck

Claude Code is impressive when it comes to:

• Understanding large codebases
• Refactoring
• Explaining logic
• Generating structured solutions

But here’s the honest truth: Sometimes Claude gets stuck in a loop.

There were moments when it kept giving me “almost correct” patches for a bug, circling around the issue without solving it. After a few rounds, I had to manually debug the problem myself—then explain the correct logic back to Claude so it could “learn” the right direction.

This taught me something important:

AI is powerful, but it still needs a human engineer to guide it.

✅ Gemini CLI: The Best for Security and Risk Awareness

When it comes to security, Gemini Coding Partner stands out.

It consistently:

• Flags risky dependencies
• Points out potential injection vulnerabilities
• Highlights unsafe patterns
• Suggests more secure alternatives

In my workflow, Gemini has become the “final gatekeeper” before code goes into production.

The Rise of Prompt Engineering: Why It Matters More Than Syntax

One of the biggest mindset shifts in AI‑assisted development is this:

Writing good prompts is becoming more important than writing perfect syntax.

A strong prompt includes:

• Clear intent
• Constraints
• Examples
• Edge cases
• Expected output format
• Context about the system

When I give AI a vague prompt, the output is vague. When I give AI a structured prompt, the output is structured.

This is why prompt engineering is not a gimmick—it’s a core engineering skill.

Vibe Coding: Let AI Write the First Draft

A concept that perfectly describes modern AI‑assisted development is “vibe coding.”

Here’s how it works:

1. You describe what you want in natural language.
2. AI generates the first draft of the code.
3. You review, refine, and correct it.
4. AI improves the code based on your feedback.
5. You finalize the architecture and logic.

This workflow has become my daily routine.

Instead of starting from a blank file, I start from an AI‑generated draft. Instead of manually writing boilerplate, I focus on design and correctness. Instead of spending hours debugging, I collaborate with AI to narrow down the issue.

This approach doesn’t replace engineering—it amplifies it.

AI Code Is Not Automatically Safe: Security Still Requires Human Judgment

One of the biggest misconceptions is that AI-generated code is “smart” enough to be secure.

It’s not.

AI can:

• Introduce vulnerabilities
• Miss edge cases
• Suggest outdated libraries
• Ignore security best practices

This is why I always run AI-generated code through:

✅ Gemini Coding Partner ✅ Static analysis tools ✅ Manual review

AI accelerates development, but security cannot be outsourced.

What AI Still Cannot Do (And Why Engineers Are Still Needed)

Despite the hype, AI cannot replace engineers. Here’s what AI still struggles with:

❌ Understanding business context

AI doesn’t know why a feature matters.

❌ Making architectural trade-offs

It cannot weigh performance vs. maintainability vs. cost.

❌ Designing systems end-to-end

It can propose patterns, but it cannot own the architecture.

❌ Handling ambiguity

AI needs clarity. Humans handle uncertainty.

❌ Taking responsibility

When something breaks, AI won’t be the one fixing production at 3 AM.

This is why the future belongs to engineers who can:

• Think critically
• Design systems
• Communicate clearly
• Guide AI effectively

The Future Engineer: A Hybrid Between Architect, Analyst, and AI Orchestrator

Based on my experience, the most valuable engineers in the AI era will be those who can:

✅ Frame problems clearly

AI needs direction.

✅ Break down complex systems

AI needs structure.

✅ Review and validate AI output

AI needs oversight.

✅ Understand security and risk

AI needs guardrails.

✅ Continuously learn new tools

AI evolves monthly.

This is not the end of software engineering. It’s the beginning of a new version of it.

My Personal AI Workflow (You Can Copy This)

Here’s the workflow I use daily:

1. Use GitHub Copilot for fast drafting

It handles the easy stuff.

2. Use Claude Code for deep reasoning

It helps with architecture, refactoring, and logic.

3. Use Gemini CLI for security checks

It catches what the others miss.

4. Perform manual review

Because responsibility still belongs to the human.

This multi‑tool approach gives me speed, clarity, and safety.

Why Continuous Learning Is Now Mandatory

AI tools evolve faster than any technology we’ve seen before.

I haven’t even had time to explore GLM 4.6, though I’m planning to test the open-source Flash version soon. With so many new models emerging, engineers must adopt a “permanent beta” mindset:

• Always learning
• Always experimenting
• Always adapting

This is the only way to stay competitive.

Practical Advice for Engineers and Students

Here are the most important takeaways from my experience:

✅ 1. Learn to write effective prompts

This is the new literacy of software engineering.

✅ 2. Let AI generate the first draft

Don’t waste time on boilerplate.

✅ 3. Always review AI output

Trust, but verify.

✅ 4. Build your own AI workflow

Different tools excel at different tasks.

✅ 5. Stay curious and keep learning

AI is evolving too fast to stay still.

Conclusion: AI Won’t Replace Engineers—But It Will Replace Engineers Who Don’t Use AI

AI is not a threat. AI is a multiplier.

• Engineers who use AI will be 10x more productive.
• Engineers who ignore AI will fall behind.
• Engineers who can direct AI will become the top 1% in the industry.

The future is not “AI vs. engineers.” The future is engineers who know how to collaborate with AI.

And I’m excited to keep exploring, experimenting, and sharing what I learn along the way.

@lerlerchan by LerLer Chan

|


|


|


|

Ler Travel Diary is using
Web Hosting and n8n on Hostinger

To be a smart saver, check out
ShopBack for more information.

Enjoy SGD5 discount voucher on
Klook with code 53E7UD

Need discount for your travels? Check out our travel deals page.