Of course, no one who’s been paying any attention to the rapid proliferation of AI throughout the enterprise needs me to tell them that this is far from the typical new-product announcement.
We are definitely in the awkward in-between phase where AI can seem like a bit of an uncontrollable animal: very powerful, somewhat unpredictable, and sometimes a little bit reckless.
It’s against that background that we find the introduction of the new Agent Commander platform today – a platform that promises not only to identify AI problems, but also to “rollback” AI-based errors.
While that’s something we’ve all been used to in other areas of IT for years, when it comes to AI security, most of us haven’t even seen this in the playbook yet.
Given that, it’s not hard to see why Veeam is getting a lot of interest in its new platform. When AI agents are making their own decisions at light speed, you’re clearly not going to be able to wait hours to detect the problem and days to fix it.
That’s been a real problem that many in security have been seeing firsthand. What’s generating the excitement around Agent Commander – and why many CTOs and other leaders in enterprise tech are getting a little excited – is that Veeam is promising to do all three: detect AI risk, protect AI workloads, and undo AI actions.
In an environment where most of our tools have been designed to detect and protect, that’s an interesting idea. Veeam’s approach is to offer security teams a single view across data, identity, and AI activities in real time.
Many existing tools do some of these tasks, but not all of them, or not in real time. I can almost imagine the debate that’s going on in CTO-level staff meetings today: “Yeah, yeah, detecting AI threats is a great thing.
But what about when an AI agent has already done something it shouldn’t do? What about when it’s already misused some sensitive data?” I’m sure many security teams have been wrestling with that question for months.
According to surveys, a high percentage of enterprises have already experienced AI-related security problems. Existing tools aren’t doing the job here. That’s why IT teams I’ve spoken to aren’t just talking about security anymore; they’re talking about trust.
They want to trust AI agents enough to continue to adopt AI for all the powerful benefits it can bring to their organizations without having to fear that their next data breach will come from their AI systems.
While Veeam’s Agent Commander isn’t a panacea, its promise of attributing AI activities and selectively reversing them could be the start of what these teams have been looking for.
But while I think that’s a big deal, I also think it’s important to take a step back and understand that this is one more sign of a larger reality about AI: We’re now officially in the age of “agentic” AI, where AI agents are not simply responding to queries and requests but are actively taking independent actions, often linked in chains of services and data.
They’re making decisions at a speed and scale that’s hard for humans to keep up with. That’s creating some brand-new risks and is why tools and frameworks designed for a previous generation of static applications look so outdated today.
Veeam’s announcement today is less a new-product launch and more of a declaration that AI is different. So where does that leave us today? I think that Veeam’s new Agent Commander is an important reminder that AI isn’t just something we’re going to have to prepare for – it’s something we’re going to have to control.
And while we’ll undoubtedly have many more questions as these autonomous AI agents continue to evolve, I think that this is a great start to the conversation. And who knows? Maybe one day we’ll actually have AI that can help us control itself.
