In July 2024, we asked a provocative question: Will AI replace the backup administrator? Read the original blog here.
At the time, Predatar had just released R16 Orca, and “Generative AI” was the phrase on everyone’s lips. “Agentic AI”– systems capable of autonomous, goal-driven action – was not yet part of mainstream thinking, at least not in the wider public domain. It wasn’t until later in 2024 that agentic AI began to enter the broader discourse. Since then, the conversation, and the tech, has moved at a blistering pace.
Our original conclusion still stands: the backup administrator is here to stay. But the nature of the role is changing far more dramatically than we anticipated. With the emergence of agentic AI, we are moving beyond assistance into true autonomy in backup and recovery operations.
So, what’s now possible? Agentic AI opens the door to capabilities such as automatically scheduling and optimizing backups based on real-time conditions, dynamically building and fine-tuning recovery plans, and intelligently orchestrating the order of restores to achieve the fastest recovery outcomes. It can also respond in real time to AI-driven anomaly detection, adjusting protection strategies or initiating defensive actions as threats emerge.
While it is possible to build these capabilities into operational management platforms, it’s still early days. The industry is just beginning to explore what fully autonomous data protection looks like, and it will be some time before these systems reach full maturity. We predict that until 2028, the model will be “human-on-the-loop”. The backup administrator role will remain as important as ever, perhaps even more so. From 2028 onwards, the role will evolve. The human backup administrator will take on more of an AI governance role. This is often described as “human-before-the-loop”, where the backup admin provides the initial guardrails that allow AI to act independently.
The threat landscape is accelerating. Backup vendors know this and are scrambling to catch up. They face a new kind of arms race against AI-enabled threat actors who are actively disrupting the ability to recover. This is no longer just about encryption or threat detection. It’s about proving resilience.
The response is clear: we must fight fire with fire. Embracing agentic AI within cyber resilience platforms is the only viable path to keep pace.
At Predatar, this shift is already well underway. We have made a decisive, seismic move to re-engineer our entire platform – spanning monitoring, automation, recovery assurance, and CleanRoom capabilities – around agentic AI. This is not a feature enhancement; it is a fundamental redesign.
Will we see a day when agents are truly autonomous, and the backup admin is fully “human-out-of-the-loop”? We’re not prepared to say. We are good, but not that good.
Stay in the loop with Predatar News
If you’re a human and you like being in the loop (from a news perspective), join our mailing list to get the latest news and insights from the Recovery Assurance trailblazers at Predatar. Sign up here.
I spend a lot of my time talking to vendors, partners and customers across the storage, security and business continuity ecosystems. Different roles, different priorities, but the same conversation keeps coming up. At some point, Clean Rooms enter the discussion and almost without fail I hear a version of the same response… “We love the idea… but Clean Rooms are expensive.”
Not All Clean Rooms Are Created Equal
Not only is the perception not universally true, but there’s another layer to it as well. Ask ten people what a “Clean Room” is and you’ll get ten different answers. For some it’s about detection, for others it’s isolation, or a last resort recovery space. The reality is not all Clean Rooms are equal, and a lot of the confusion in the market comes from that.
The Industry Created This Problem
To be fair, the perception around cost didn’t come from nowhere. The industry has, over time, shaped it. The industry has tied Clean Rooms to heavy, hardware led approaches, large deployments, dedicated infrastructure and significant upfront investment before you see any value.
Traditional solutions such as the Dell CyberRecovery Vault is a good example of where that thinking comes from. They are built around sizeable hardware estates and focus on encryption detection at scale, identifying when ransomware has started encrypting data. That has real value, but it also tells you something important. By the time you are detecting encryption, the attack is already in motion. You are reacting, not proving anything and you have likely already invested heavily just to get there.
Costs Are Making It Worse
Fast forward to 2026 and the situation is getting tougher. Hardware costs have jumped significantly, with increases of up to 50% not uncommon. So the mental equation becomes even harder to justify. People see Clean Rooms as valuable, but expensive, complex, and something to come back to later.
But that is not what a Clean Room should be.
The Only Question That Matters
When I speak to customers, especially security and business continuity teams, very few are asking for better detection tools. What they really want to know is much simpler.
“If we get hit, can we recover? Not in theory, not based on a report, but for real. Can we recover quickly, cleanly, and completely?”
That is the problem that matters, and it is the problem we set out to solve with Predatar CleanRoomTM, our Isolated Recovery Environment.
Proving Recovery Changes Everything
Instead of waiting for something bad to happen, the Predatar approach is to continuously prove that recovery works. We automatically recover backups and primary snapshots into isolation, test whether those systems come back clean and usable and scan for malware as part of that process. You do not need any manual effort to kick this off. We run fully automated restore testing at scale across both primary and secondary data and we do it in a completely vendor agnostic way.
That combination matters. It means you are not tied to a single backup platform or storage vendor, and you are not relying on periodic, manual testing that may or may not happen. You are continuously validating your ability to recover across your actual estate.
What You Find When You Actually Test
And when organisations start doing this properly, something interesting happens. They do not just gain confidence in recovery, they uncover risks they did not know existed. Dormant malware, planted threats, things that have not triggered yet but would have eventually. Across our customer base, 93% have found malware sitting in their backups. These are not organisations in the middle of an attack. These are organisations doing the right thing, testing and validating and still finding hidden exposure.
Why Cost Still Dominates the Conversation
So why does cost still dominate the conversation? Because most people still believe Clean Rooms require a huge upfront commitment. Big hardware, big rollout, big decision. And when something feels that significant, it gets delayed. I see it all the time, projects pushed into next year, endless scoping exercises, teams waiting for the right moment. But resilience doesn’t work like that. You do not get to choose when you need it.
A Clean Room Without the Hardware Burden
This is exactly why we took a different approach with Predatar. We deliver the Predatar CleanRoom as a virtual appliance. You can deploy it into your existing environment, whether that is your virtual platform, repurposed hardware or simply spare capacity you already have. In fact, over 70% of our customers deploy using resources they already own, with no new hardware investment required. What used to be a capital heavy decision becomes an operational one, and that changes how organisations approach it.
Start Small and Scale When You’re Ready
Many teams assume that if they adopt a Clean Room, they have to do everything at once. In reality, the most successful organisations start small. They focus on what matters most, a minimum viable business service, a critical application, or a subset of their environment such as key VMware workloads. They prove recovery there first and then expand over time. There is no need for a big bang rollout or an all in commitment from day one.
One View Across a Fragmented Estate
Another challenge many organisations face is fragmentation. Multiple backup platforms, multiple storage systems, and multiple tools all trying to answer the same question in slightly different ways. Can we recover? Predatar works across backups and primary snapshots, across vendors and platforms, to give you a single, consistent view of recovery confidence.
The Real Barrier Isn’t Technology
If there is one thing I have learned from these conversations, it is this: perception, not technology, is the biggest barrier to adopting Clean Rooms. If people believe they are expensive, complex, and hardware-heavy, they will keep putting them off. But when we redefine them as something lightweight, flexible, and focused on proving recovery, the conversation changes very quickly.
It All Comes Down to One Question
Because in the end, resilience is not about how fast you detect a problem. It is about what happens next. Can you recover quickly, cleanly, and completely?
That is the only question that really matters, and it is one the industry needs to get better at answering.
If you are having the same conversations and want to challenge the assumptions around Clean Rooms, we would welcome the discussion. Drop us a line.
If you’re a risk or compliance officer – or, in fact, anyone with a role to play in ensuring operational resilience in your organization – how do you know that the backup platforms your IT team is using are up to the task?
It’s an important question. The answer – when you think about it – is worrying.
Because typically, organizations like yours rely on the resiliency credentials that the technology vendors themselves set out. Is it any wonder that every backup vendor claims that their platform is the most resilient, when they are setting their own criteria to measure resilience?
Of course, you could ask your IT or infrastructure team – they are the experts, after all. But the brutal truth is that resilience isn’t their top priority. These teams are typically overstretched and under-resourced. Making sure that the systems and data your organization runs on every day are available under normal conditions keeps them busy enough.
Understandably, they will have a preference for the infrastructure products that make everyday management and administration easy. And, in fact, it goes beyond day-to-day practicalities. If you spend enough time with storage teams, you’ll notice that people often become deeply attached to certain vendors, defending them almost like a belief system rather than just a technology choice.
Understanding the true picture of resilience in your business becomes even more complicated when you have a large and complex infrastructure environment. An environment made up of technologies from multiple vendors – each with its own claims, its own standards, and its own devoted followers inside your business.
The idea of ‘devoted followers’ brings an interesting concept to mind – one which history has shown to be a good one: the separation of church and state. Not only has this approach been useful for monarchs who have found themselves inconveniently married, but more importantly, it has been invaluable for managing kingdoms. And, believe it or not, the concept is relevant here.
In this analogy, vendors like Dell, IBM, Pure, Veeam, Rubrik, etc are the ‘churches,’ each with its own doctrine. Useful, certainly – but not neutral.
What is needed is a ‘state’: a vendor-independent authority that sits above the fray, governing consistently across all systems. One that ensures recovery testing is meaningful, malware scanning is credible, and resilience is measured the same way everywhere.
This separation reduces conflicts of interest, standardises outcomes, and brings a much-needed layer of objectivity to an otherwise fragmented landscape.
No analogy is perfect, of course. Vendors aren’t religions, and governance platforms don’t wield civil authority, but the underlying principle still holds: resilience shouldn’t be defined by the tools themselves. Because if it is, don’t be surprised if everything looks perfect… until it really matters… and then it isn’t.
Discover Independent Recovery Assurance
Visit www.predatar.com to start your journey to independent Recovery Assurance or contact hello@predatar.com for expert advice from the leading independent Recovery Assurance experts at Predatar.
R18.2 reinforces Predatar’s core mission to deliver proof. Proof that your backups recover quickly, cleanly, and completely across your entire estate. Not just within a single backup product, but across primary storage and multiple backup platforms working together to restore real business services when it matters most.
This release adds support for NovaStor Backup, the latest addition to Predatar’s growing ecosystem of supported vendors. With this integration, teams can orchestrate recovery testing end-to-end, validating not just individual restores, but true recovery by bringing together data from multiple sources to fully re-establish services.
The Predatar Cleanroom scans recovered data for malware in an isolated environment, helping ensure it is safe, compliant, and ready for use. Governance reporting delivers clear, auditable insight into every test, giving organisations the confidence to demonstrate resilience across their entire data landscape.
NovaStor joins a broad and growing set of supported platforms, including Cohesity, IBM Storage Protect, IBM Storage Protect+, IBM Data Protect, Rubrik, Veeam, Zerto, Pure Storage, and IBM FlashSystem, so you can prove recovery across the technologies that underpin your business.
Stefan Utzinger, CEO of NovaStor, commented:
“NovaStor customers can now extend their data protection strategy with automated recovery testing and cleanroom validation. Together with Predatar, they gain the assurance that their backups don’t just exist, they are proven to recover in a way that supports real business continuity.”
At its core, Predatar goes beyond restore. It proves recovery.
To learn more about Recovery Assurance for NovaStor backups, see the latest release information on the Predatar website.
