As enterprises rush to deploy AI, a new problem is emerging: how to stay in control once machines start making decisions on their own.
The rise of agentic AI—systems capable of acting autonomously across applications, data, and infrastructure—is shifting the risk landscape dramatically. While the promise of automation and productivity gains is significant, so too is the potential for unintended consequences.
Commvault’s latest announcement reflects this shift, introducing a set of capabilities designed not just to enable AI, but to govern and recover it.
At the heart of the challenge is trust. AI systems are only as reliable as the data they are built on—and increasingly, that data is distributed across hybrid environments, legacy systems, and cloud platforms.
“AI is becoming a system of record for the enterprise,” said CEO Sanjay Mirchandani. “But if the underlying data is compromised, the AI itself becomes unreliable.”
This is where the concept of “AI resilience” comes into play.
Rather than treating AI as a standalone capability, organisations are being forced to think about it as part of a broader operational resilience strategy—one that includes governance, observability, and the ability to roll back changes when things go wrong.
The introduction of tools like Data Activate and AI Protect reflects a growing recognition that AI pipelines must be tightly controlled. From curating trusted datasets to mapping agent activity and enabling full-stack recovery, the focus is shifting from experimentation to operational discipline.
This is particularly relevant in highly regulated markets such as Australia, where frameworks like APRA CPS 230 and CPS 234 are raising the bar for operational resilience and data security.
In these environments, it’s no longer enough to deploy AI—it must be auditable, governable, and recoverable.
The emergence of platforms that combine AI enablement with resilience capabilities signals a broader trend: the convergence of AI, security, and data governance into a single operational layer.
As agentic AI continues to evolve, organisations that can maintain control over their data—and recover quickly when things go wrong—will be best positioned to scale safely.
