Restoring Data Isn't Enough In The Age Of AI Agents

Commvault Field CTO Vidya Shankaran explains why AI agents and increasingly complex data environments are forcing IT leaders to rethink what cyber recovery really means.

(Vidya Shankaran, Field CTO – Head of Americas, Commvault)

Restoring data after a disaster used to be a fairly straightforward process: Find the clean recovery pont, restore the data, and you're back in business (assuming the backup or snapshot was good!)

Now, AI agents are making that process considerably more complicated.

Agents are, of course, increasingly moving into production environments. They often have access to enterprise data, permissions, configurations and workflows.

IT leaders now have to consider not only what happens when data is compromised, but what if the AI agent is compromised as well?

That emerging resilience challenge is one of several topics Vidya Shankaran, field CTO for cloud security and emerging technology at Commvault, discussed on MES' Ready.Set.Midmarket! podcast.

Shankaran argues that IT leaders should distinguish between restoration and recovery.

As she explained, restoring data to a clean point does not mean the business is fully recovered and can operate safely again.

AI complicates matters. If an AI agent is compromised, organizations may need to consider the interactions between the agent and the enterprise data and how to restore both to a clean state without disrupting the rest of the environment.

Shankaran says too many organizations still approach recovery as paper-based, checkbox exercise. Recovery should instead be tested in clean-room environments that can help expose process gaps and give teams experience in recovering before they are under pressure following a real incident.

Watch the full episode for a deeper dive into agents, identity, clean-room testing, and what midmarket IT teams should be considering as recovery and resilience become more complicated.