Quality can’t just be a vibe: The new QA mandate in the age of AI (opinion article at Executive Digest)
By Guilherme Ramos Pereira, Head of Strategic Development at 99x Portugal (original article in Portuguese published at Executive Digest)
Lately, I’ve been reflecting a lot on how Artificial Intelligence (AI) has moved from being a distant promise to becoming a reality that’s reshaping everything: business models, internal processes, and even the way we work every day. In the middle of this accelerated transformation, one thing has become increasingly clear to me: Quality Assurance (QA) is no longer just a “technical step” in the software development cycle. It has gained a strategic, almost transversal dimension, because it is what ensures that every AI-driven decision is truly reliable and consistent.
It’s true that QA teams have long been involved from the early requirements stage. Methodologies such as Behavior Driven Development are proof of that. But the mass arrival of AI has changed the game. Validating software “the old way” is no longer enough: it’s now essential to track, in parallel, the business logic, the data feeding the models, and the processes sustaining every automated decision. When AI accelerates code generation like never before, a reactive approach is not just insufficient—it becomes dangerous.
A study by MIT Technology Review Insights shows that 95% of companies are already integrating AI technologies into their workflows. They use them to automate tasks, analyze data, and create solutions faster. But most are still in pilot stage. And here lies the risk: the proliferation of MVPs created through clever prompts and a fast-paced “vibe coding” culture encourages experimentation, but doesn’t always ensure security or alignment with strategy. If requirements are incomplete or poorly defined, AI fills the gaps with assumptions. And we know where that can lead: bugs, logic errors, and products misaligned with real objectives.
In a conversation with João Cardoso, a long-time colleague and digital transformation expert, we reached a simple but unavoidable conclusion: the hype of speed cannot crush the most basic quality processes. Rushing can be the enemy of viability. And perhaps that’s why it’s urgent to embrace a new “mandate” for Quality.
Today, QA is not just about code. It also looks at requirements, edge cases, data integrity, and consistency when different modules are integrated. This is where we prevent the domino effect: small flaws at the start that turn into major issues later on. Which is why the solution goes far beyond smart testing. It requires redesigning workflows and validating every stage, from data collection all the way to a model going live in production.
AI brings speed, scale, and coverage. Fantastic, no doubt. But humans bring something no machine can replace: context, creativity, and strategic vision. Balance only emerges with clear processes, well-defined prompts, and continuous checks. This balance is what safeguards the value chain and builds trust in the outcomes.
The data backs it up: according to MIT, 45% of executives see governance, security, and privacy as barriers, and half consider poor data quality the biggest obstacle. In fact, 98% of companies say they’d rather forgo being pioneers in AI adoption if it means achieving safer, more reliable implementation.
Ultimately, the real question isn’t “what can we do with AI?” but rather “what can we do well—with confidence and real impact?” And the answer inevitably lies in this new mandate for Quality. Not as an isolated phase at the end of the line, but as a guiding principle throughout the entire journey. Uniting AI, data, and people to build decisions that are solid, consistent, and transformative.
