Software runs the world, and how it’s built is ever-changing. However, that doesn’t mean it’s always built well—or as good as it can be.
Today, the stakes are higher than ever before. AI is helping to crank out new software and applications faster than ever, but the “shift-left” movement that gathered steam more than a decade ago is failing to make sure those created applications are actually being adequately tested to ensure they work as intended.
Shift left put the emphasis on developers to test software earlier in the development life cycle. The idea was sound; earlier testing would catch problems sooner and improve quality. In practice, though, the approach was often poorly implemented. Over time, it led to a de-emphasis of dedicated testing disciplines rather than a true elevation of quality across the software development life cycle.PROMOTEDhttps://d7be09101f7ea7a50a1821d94a2102de.safeframe.googlesyndication.com/safeframe/1-0-45/html/container.html
Now, AI is magnifying the impact of that lost focus on testing. Our 2026 report found that 60% of software and quality assurance experts we surveyed experienced quality issues in the past year. Almost seven of 10 said quality has already diminished and will slide even more in the next year.
That’s not good odds for entities that rely on software applications to drive revenue and deliver customer service or a government service. One application failure can be devastating both to the application owner and the application user. A 2022 report published by the Consortium for Information & Software Quality estimated that poor software quality was costing $2.4 trillion annually in the U.S. alone.
Yet while AI copilots generate code faster than ever, traditional quality assurance processes still rely heavily on manual effort, linear workflows and outdated success metrics. Our research found that almost 60% of teams do more than 40% of their application testing manually. At the same time, AI is producing or helping to produce more software code.
A Building Quality Crisis
The gap between how quickly software is created and how thoroughly it is validated is widening—and that gap is unsustainable.
That’s why “shift-left”-focused testing environments need to change course. Even if developers test code to make sure it’s free of bugs and works on its own, that doesn’t mean the resulting application will work as intended. Clean code doesn’t entail an effective end-user experience. When code gets compiled into applications—and is then required to work under high demand—stuff can break. That stuff is what a truly rigorous application testing approach would catch.
No doubt, autonomous testing that can keep pace with autonomous code creation will be essential to close the gap. However, that’s only part of the solution.
A Mindset Change
The deeper shift must happen higher up the organization in terms of mindset. Software development can’t continue to be treated primarily as a developer problem. Business priorities must govern it first and foremost. Is speed to market more important than how resilient the application is when it gets there? What’s the true cost of an application failure—not only in the moment but also in a long tail of potential damage to brand reputation?
The software development life cycle has shifted before, demonstrating the industry’s ability to adapt. The waterfall era of the 1970s, for instance, treated software like physical infrastructure: rigid, sequential and expensive to change. Decades later, agile reframed software as a living product, emphasizing adaptability and customer feedback. DevOps followed, pushing for faster releases and fewer defects, with shift-left testing becoming mainstream. Cloud-native development then enabled applications designed for constant evolution.
Competitive pressure will force the industry to change again.
When one company launches a promotion and its application fails—triggering downtime, lost revenue and brand damage—while a competitor succeeds because it invested in adequate application testing to keep up with AI-accelerated software development, the lesson becomes unmistakable. The companies that rethink the software development life cycle first will be the ones that win.
Here’s how to be one of them:
• Shift from “fast” to “resilient.” Speed to release matters, but reliability and security matter more. An insecure or unstable application is a liability, not an asset.
• Get honest about risk. Leaders must understand the real business impact of applications that crash and disappoint.
• Shift right—continuously. Early testing is necessary, but it’s not sufficient. Post-deployment monitoring and validation are critical to ensuring applications behave as expected in production.
• Redefine success metrics. Success isn’t simply “Does it work?” It’s “Does it deliver the intended business outcome?”
• Upskill QA teams. Quality professionals need a strong understanding of AI so they can work alongside it—leveraging its strengths while validating its outputs.
• Arm QA with autonomy. Manual testing alone can’t keep up. Autonomous quality testing is required to match the speed and scale of AI-driven development.
The industry today isn’t provisioned to handle the volume of applications and constant change that AI is unleashing. That reality will force a reckoning.
Software development is a life cycle. From inception to deployment to ongoing operation, the central question remains the same: Does this software work—and does it work for the business? The organizations that reimagine this life cycle the fastest will define the next era of competitive advantage.
https://www.forbes.com/councils/forbestechcouncil/2026/03/31/why-shift-left-is-breaking-down-in-the-age-of-ai/a>