From:
Babu Manickam
Chennai, India
Dear fellow tester,
My name is Babu Manickam. I’ve been in quality engineering for over 27 years.
I’ve trained more than 50,000 professionals across 32 countries. Built testing solutions for healthcare, banking, e-commerce, and enterprise clients.
I run two companies — Testleaf and QEagle — and I’ve spent my entire career helping testers get better at what they do.
But I’m not writing to you today to talk about any of that.
I’m writing because something happened in December 2025 that completely changed how I think about AI, Copilot, and what’s possible for testers like you and me.
And I need to tell you the story…
I Thought I Was Going to 100% Fail This Project
In early December 2025, a medical customer from Illinois, USA, approached us with an urgent request…
They needed a full Selenium-to-Playwright migration. Hundreds of test cases. Complex workflows. Healthcare compliance requirements.
The competition was fierce. Other vendors had already quoted lower prices.
The deadline? 6 weeks.
And here’s the thing…
I said yes before I had a team ready.
“What was I thinking? I don’t have a bench team with strong Playwright skills. The pricing is barely profitable. And I just committed to a 6-week deadline with a US healthcare client…”
The first two weeks were brutal.
Manual migration was eating all our time. My team was stuck converting test after test, line by line.
I was working 14-hour days and we were still falling behind.
I couldn’t sleep. I kept running the numbers in my head.
At this pace, we’d miss the deadline by at least few weeks.
I started thinking about how to break the news to the client. How to explain that we needed more time. How to not lose the contract entirely…
Me, pretending everything was fine during a client status call
Then I Stopped Using Copilot Like Everyone Else…
Up until that point, I was using GitHub Copilot like most developers do — autocomplete on steroids. Tab, accept, move on.
But one late night, out of desperation, I started experimenting with something different…
Instead of using Copilot as a coding assistant, I started treating it like a team of focused, specialized agents.
I wrote custom instruction files that encoded our testing standards.
I created agent skills for repetitive migration patterns. I connected Copilot to our test management system through an MCP server.
And something incredible happened…
Work that was taking my team 4 hours per test suite started taking 45 minutes.
Not because the code was sloppy. Not because we cut corners.
Because Copilot was now working WITH our standards, not against them.
Within days, we were back on schedule. Within two weeks, we were AHEAD.
We Delivered. On Time. Under Budget.
The client was thrilled. They extended the contract.
My team couldn’t believe the difference.
And I sat there, staring at my screen, thinking…
“Why isn’t every tester using Copilot this way?”
That Exact Approach — Is What I’m Going to Teach You in 120 Minutes
Every technique I used to save that project… every Copilot feature I discovered… every shortcut that turned weeks of work into days…
I’m going to walk you through it. Live. Step by step.
Not a slide deck. Not a pre-recorded tutorial. A live 120-minute session where I build an AI agent from scratch using the exact same techniques.
And the best part?
You don’t need to be an AI expert. You don’t need ML knowledge. You don’t even need to be a strong coder.
If you can write a test case, you can do this.
But first, let me share the 3 things I learned that changed everything…
