AI Coding Creates a Verification Gap
Mike Vizard speaks with Prince Kohli, CEO of Sauce Labs, about AI software verification and the growing gap between faster code generation and slower testing processes. Kohli says AI coding tools can help teams write code much faster, but verification has not improved at the same rate.

That imbalance is creating a new software quality challenge. Kohli explains that many teams are seeing code creation increase dramatically, while release velocity improves only modestly. Bugs are also rising, especially when AI-generated code moves through older testing processes that were not built for this speed.
Code Velocity Is Not Product Velocity
The discussion highlights a key distinction for software leaders. More code does not always mean more product value. Kohli says organizations often confuse code velocity with product velocity. AI tools can generate applications quickly, but teams still need reviews, test authoring, device coverage, user journey validation and production-like verification.

AI software verification becomes even more important when developers are reading more code than they write. They may not fully understand the AI-generated architecture or the context behind each change. That makes traditional code review and happy-path testing less reliable. It also increases the chance that subtle defects reach production.
Testing Needs an Independent Check
Kohli warns against using the same AI model to write code and verify that code. He compares it to asking AI to grade its own homework. Mature organizations are taking a different path. They are using independent testing systems that evaluate application intent, author tests, run them in the right environments and troubleshoot failures.

This approach gives teams a second check on AI-generated software. It also helps testing keep pace with code creation. Sauce Labs is positioning AI software verification as a way to close that gap without slowing innovation or abandoning AI coding tools.
Production Risk Is Already Showing Up
Kohli also shares findings from Sauce Labs research. He says 80% of organizations traced a production incident or outage to AI-generated code. He also notes that 90% reported serious business impact, while 66% admitted they compromised quality or testing standards to meet faster release deadlines.

The practical advice is not to avoid AI. Kohli says teams should use AI to improve software delivery, but they must invest in verification at the same time. AI software verification gives organizations a path to move faster, reduce production risk and protect software quality as AI coding becomes more common.