What one performance engineering leader would tell industry newcomers who are worried about AI

Quick summary: AI is creating anxiety and excitement — teams can get more work done faster, but does all this automation leave the worker behind? Not necessarily, says one performance engineering leader. The AI revolution, he says, is another technological wave. To ride it, performance engineers must embrace the change.

AI: Friend or foe?

Most people who work in tech see AI as some combination of scary and exciting. It can do a lot of easy, low-grade work for you, but it gets sporadically better at some tasks over time, and it’s not always clear where that leaves the people behind the work.

Software development is the first field it seems intent on conquering. Software is being built faster, there’s more of it, and AI tools have made their way into the testing space now, too.

One way to get some perspective is to zoom out. Technology has shifted like this before, though maybe not as fast or visibly. In performance engineering, the changes in the last couple of decades offer a surprisingly helpful lens.

Mike Hawkins, Solution Architect at Tricentis, has seen a lot of change. He’s been working in performance engineering for more than 20 years and says that watching the field’s evolution firsthand has made him, if not calm about AI, then at least a little less alarmed.

Take the long view

Hawkins’ career is firmly planted in the performance engineering space — it’s what he does. But he started his career in the U.S. Navy working in cryptology and early IT. From there, he moved into systems work and eventually into performance validation in the late 1990s, just as the discipline was beginning to take shape.

He taught himself to code — first in VB, then in ANSI C — and built his career in an era when performance testing entailed writing and maintaining large volumes of scripts by hand.

Conference highlights: How to performance test in minutes and get results faster using a SaaS based approach

Later, he moved into consulting, where he helped large companies assess and rebuild their performance testing practices. That meant everything from evaluating tools to designing frameworks and training teams that would eventually inherit the work.

Looking back, the limitations of early performance testing were pretty clear, he says. Tests were narrow and repetitive, often hammering a system with the same request again and again. There wasn’t much of a concept of real-world simulation.

“You’re not exercising the system like it needs to be exercised,” Hawkins says. “It’s one thing to have a loop where Bob Smith is purchasing a pair of shoes over and over and over. That’s not how real users behave.”

Testing didn’t allow for variations like different users, different data, and different workflows. Today’s tools do a much better job of capturing that complexity, simulating concurrency, and modeling how systems actually behave under pressure, he says.

The fundamentals haven’t changed much, though.

“You’re still dealing with CPU, memory, disk I/O, and network,” Hawkins says. “Those principles haven’t changed since the beginning of performance engineering.”

Hear more from Mike at our webinar: 3 MCP use cases that demonstrate NeoLoad’s latest agentic capabilities

Change is constant (but deeply unpleasant)

For many early-career tech workers, AI feels disruptive. But it’s not the first time Mike has felt the ground shift.

Years ago, when no-code and low-code testing tools started gaining traction, he was more than a little skeptical. A UI didn’t seem like it could give him the flexibility and control he had with code.

But at the core, some of his skepticism was personal.

“I had a skill set that I was pretty darn good at, and now here’s this new way of doing it that makes my skillset less valuable,” he says.

Then he took a job that tasked him with implementing one of those tools, Tricentis NeoLoad, on a real project. The results were hard to argue with. He saw script turnaround drop from weeks to days and error rates fall from roughly 10% to under 0.5%. Releases stopped slipping entirely.

Mike would join Tricentis in 2022 to work on NeoLoad.

We’re going to need a bigger surfboard

AI feels like an even bigger change than the move to no-code performance validation, and it’s complicated by the scale of code being generated. Depending on who you ask (and maybe which investors are present) 30% to 100% of an organization’s code can be written by AI.

If the anxiety felt by some is that performance engineering as a profession will be replaced by AI, Hawkins says that idea seems unlikely and a little dangerous.

“You would be setting yourself up for incredible risk,” he says.

If software developers are producing more code faster, companies will need testing to move much faster. That means performance engineers need to be experts at performance engineering as well as AI tooling.

Even in a “shift left” world where developers can use AI to run basic performance validations as soon as they write code, someone still has to test the full, end-to-end behavior of a system under real-world conditions.

“You’re always the last rung in the ladder,” Hawkins says. “The last one to get the code and to get an environment that works.”

Rather than seeing AI as a potential contender for your role, consider it the tool that helps you keep pace with rapid release cycles.

eBook: Early insights, better outcomes: Shifting left with observability

In the future, Hawkins says he sees AI-fluent performance engineers as valuable assets to software teams. For newer entrants, it’s trickier. They’re being asked to learn both the fundamentals of their field and how to effectively collaborate with AI at the same time — they need to develop both skillsets to stay competitive.

Still, Hawkins says it’s much more likely that this is another change, albeit a big one, to the industry rather than an all-out elimination of the need for human oversight. To him, the AI revolution is yet another technology wave to ride.

“You’re always looking for that big wave,” he says. “When you see it coming, prepare yourself and then jump on it and ride that. But always be scanning the horizon for the next one.”