Artificial intelligence is rewriting the rules of software development. The headlines tell us developers are the ones on the front lines – augmented by AI to code faster, ship faster, and in some cases, even build billion-dollar companies singlehandedly. But there’s another, quieter transformation happening in parallel, and it might shape the next era of software more than people realize.
It’s unfolding in quality assurance.
Quality as the counterweight to velocity
In business terms, software quality is the other side of the coin to development velocity. You can produce code at lightning speed, but if it doesn’t meet your quality bar, you’ve built nothing of lasting value. In fact, the faster development teams go, the more pressure there is on QA to keep up.
Writing code was never the bottleneck. I might spend two hours thinking about a change – structuring it, refactoring, breaking the problem down – before maybe only spending 12 minutes actually writing it. GitHub Copilot can write that code in 30 seconds, but that’s not the hard part.
That thinking process, where you break the problem down and identify important relationships between components, deduplicating (which AI is still not very good at doing), these help determine whether a feature works as intended and aligns with the product’s vision. Now, with AI accelerating the coding phase, the volume of changes hitting QA is increasing dramatically. More features, more changes, and more opportunities for errors to slip through all arrive faster than before.
When quality can match this new speed without compromising standards, the business can move faster overall. When it can’t, velocity grinds to a halt – not because the developers slowed down, but because the downstream checks became the new bottleneck.
That sets up an important question: If QA can finally keep pace, what more could it accomplish beyond its traditional guardrails?
From “checking the boxes” to shaping the product
AI is expanding QA’s role far beyond verifying that features work. Traditionally, automated testing has focused on functional checks:
Does this button trigger the right event? Does the API return the expected response? Does the form validate correctly?
Those are still essential – but AI now makes it possible to programmatically assess aspects of the product that used to require manual review.
There are two major shifts here.
Programmatic qualitative checks: You can take a screenshot of your application and have an AI model determine if the buttons are in the right place, if the color scheme is consistent, or if the layout is intuitive. A manual tester could answer those questions, but they’d be doing it at the speed of a human. Now it can be done at the speed of code.
That means automated testing can include qualitative, subjective determinations at scale – something that simply wasn’t possible before. Instead of splitting automation and exploratory/manual testing into separate worlds, they can now merge into a more unified, powerful process aimed at holistic application integrity.
Unlocking new automated tests through structured data: The second shift is about expanding the scope of what QA even bothers to check. In the past, certain tests weren’t worth doing because preparing the data was too time-consuming. During a web session, you can now capture API calls made by the application. An AI model can instantly convert that pseudo-structured capture into reusable structured data, which you can feed back into a test harness to independently validate the API.
Before, the ROI of building that system just wasn’t there. Now it’s easy and free to do, so there’s no reason not to. Even if those tests only occasionally catch an issue, their low cost means they’re worth running – and in that rare “once in a century” bug scenario, the payoff can be huge.
Both of these capabilities push QA into more strategic territory. Instead of only certifying that features work, testers can influence the product’s usability, consistency, and overall customer experience at a much larger scale.
That larger influence brings with it a new responsibility: Making sure those insights reach beyond QA and into the hands of the people who can act on them.
Communicating beyond QA
With this expanded scope comes a new challenge:
Communicating the value of these insights to stakeholders outside the QA team.
Identifying a UI flaw or performance issue is one thing; framing it in a way that resonates with product managers, designers, or executives is another.
Not everyone’s professional mindset is mission-aligned with their organization. The CTO might want to hear how this reduces bottlenecks and frees up resources, while a tester might be more motivated by how it expands their influence and skill set.
The most effective testers learn to tailor their message to the person receiving it. That means understanding not only what you want to communicate, but also how that audience prefers to receive information. If the design lead expects feedback to come in a certain format or through a particular channel, following that protocol increases the likelihood that your insight will be acted on.
Of course, AI can even assist here, drafting communications in the voice, style, and format appropriate for each role – whether that’s a concise technical report for an engineering lead, a business-impact summary for an executive, or a usability narrative for a product owner.
When testers can combine technical observation with strong cross-functional communication skills, they transform from bug finders into trusted advisors who can influence both the product roadmap and the business strategy.
And that’s a critical skill to have, because the expectations for what QA can deliver are about to rise sharply.
The new table stakes
The urgency isn’t about AI taking a tester’s current job. It’s about the next job. As automated, AI-powered checks become the norm, the expectations for what QA can deliver will rise. The ability to design and interpret these tests will become table stakes.
It’s a shift I compare to the evolution of internet bandwidth. When Skype first emerged, people complained about poor call quality and said we just needed DSL, and it would be fine. Then DSL arrived, and suddenly new services like YouTube pushed demand even higher. No matter how much capacity you have, people will find ways to fill it if there’s utility.
The same will happen in QA. As soon as you can automatically test more aspects of a product – UI layouts, performance under varying conditions, API responsiveness – teams will find ways to depend on those capabilities. The development lifecycle of the future will assume such speed and coverage, and organizations that embrace this perspective will move faster, win markets, and set new standards.
You can keep doing what you’re doing, but that’s what you’ll be doing forever in a fixed-size pool. The next opportunity (i.e., growth!) won’t be there unless you adapt.
For those ready to adapt, there’s more than just job security at stake – there’s a real opportunity to lead.
Fitz Nowlan is VP of AI and Architecture at SmartBear.
