Veeam Software Group GmbH used VeeamON 2026 in New York City this week to punctuate its shift from “the backup company” to a data and artificial intelligence trust platform for the agentic era.

With a new architectural layer and an aggressive product roadmap, Chief Executive Anand Eswaran (pictured) and President of Products and Technology Rehan Jalil are betting that the next decade of enterprise infrastructure will be defined less by how quickly you can restore a virtual machine or data set and more by how confidently you can let AI act on your data.

From backup vendor to trust layer

For most of its 20-year history, Veeam has been synonymous with backups and fast recovery, to the point that “instant recovery” became part of the company’s identity. Eswaran reminded the VeeamON audience that Veeam earned its leadership by reducing customers’ RTOs from hours to about two minutes and by building “the broadest workload coverage on the planet across virtual machines, physical, hybrid multicloud and SaaS.” That foundation now protects more than 550,000 customers in more than 150 countries, including 82 of the Fortune 500, and drives more than $2 billion in annual recurring revenue.

In Eswaran’s view, what has changed is not the importance of recovery but the nature of the threats and the actors who access enterprise data. He framed Veeam’s history as three eras: traditional backup and recovery (“assume restore”), cyber resilience (“assume breach”), and now the agentic era of AI (“assume autonomy”), in which nonhuman identities operate at a scale and speed that existing tools were never designed to govern.

Defining the agentic AI problem

At the core of Veeam’s pivot is a view of how AI is deployed across large enterprises. Veeam’s research and telemetry indicate that autonomous AI agents already outnumber human employees by 82 to 1 on average, representing more than 250,000 non-human identities per organization. Even more concerning, Veeam reports that 97% of those agents have excessive privileges, dramatically expanding the blast radius of a single compromised or misconfigured agent.

Eswaran argued that legacy security architectures implicitly assumed “the actor was human,” and that assumption “just fundamentally broke instantly” when agents began accessing ERP, CRM, warehouses, email, files and SaaS systems in parallel. In that world, a new failure is “not just a breach, it’s a wrong decision executed at machine speed before anyone notices” — a point underscored by examples such as an AI agent deleting a production database and its backups in nine seconds, or autonomously recreating a cloud environment and triggering a 13-hour outage and millions of lost orders.

The missing layer in the AI stack

Veeam claims there is now a “missing layer” in the AI stack, positioned between data platforms and models. The solution is a unified data and AI trust layer that treats data, identities, access, regulatory posture and resilience as a single system. “The infrastructure to deploy AI exists,” Eswaran told attendees. “The infrastructure to trust it doesn’t.”

That thesis is why the company built its new Veeam DataAI Command Platform, described as “the industry’s first unified data and AI trust infrastructure for the agentic era.” The platform is the result of Veeam’s December 2025 acquisition of Securiti, a leading data and AI security posture management vendor, combined with two decades of Veeam’s recovery and data protection capabilities. By design, it spans both production and backup systems, enabling visibility into what AI agents can access, what they did and how to undo it with precision.

Inside the DataAI Command Platform

Architecturally, the DataAI Command Platform is anchored by the DataAI Command Graph, which Veeam calls a unified intelligence layer with more than 300 connectors spanning public cloud services, SaaS applications, on-premises systems, and now backup environments. Jalil described it as a “social graph for data,” continuously mapping data assets, users, permissions, AI agents, activity and protection status across billions of files and millions of tables.

On top of that graph, Veeam has defined six integrated capabilities as the core of its trust layer:

• DataAI Security combines data security posture management, identity intelligence and detection of toxic combinations, such as sensitive data exposed to overprivileged agents or external users.
• DataAI Governance enforces controls at the data source rather than at the agent, so both sanctioned and shadow agents “hit a brick wall” when they try to access governed sensitive data.
• DataAI Compliance maps to more than 100 regulatory frameworks, including the EU AI Act, DORA, GDPR, HIPAA, NIST and AI RMF, generating “auditable and defensible proof” on demand for boards and regulators.
• DataAI Privacy, powered by a “People Data Graph” that unifies structured and unstructured personal data, automates consent, data subject rights, data minimization and cross-border transfer controls in real time.
• DataAI Precision Resilience uses the graph’s deep context to “undo exactly what went wrong without rewinding the entire system,” down to a specific data element or five seconds of agent activity.

An agentic layer of built-in AI assistants can answer natural-language questions, such as “Is workload X protected?” and automate tasks like log triage, ticket management, and policy-driven recovery. Jalil explained the reality of the situation: “If you don’t understand what data you have, who’s touching it, and what changed, there is no automation, no precision and no compliance.”

Product proof points for the pivot

The 2025 edition of VeeamON marked the moment when the company outlined its vision, and this year, it focused on releasing products that align with the trust narrative. The headline announcement was the previously mentioned Veeam DataAI Command Platform, positioned as the missing AI trust layer and immediately available with the DataAI Command Graph and five core domains live. Existing Veeam Data Platform customers can connect to it via a new DataAI Resilience Module, gaining centralized visibility and agentic capabilities “with no re-migration required.”

On the resilience front, Veeam previewed Veeam Data Platform v13.1, introducing more than 70 features, including expanded hypervisor coverage (targeting 95%+ of the market), portability across hypervisors, stronger Active Directory Forest recovery, post-quantum cryptography enhancements and smarter NAS archiving for lower-cost long-term retention.

The company also introduced Veeam Intelligent ResOps, the first resilience offering built natively on the DataAI Command Platform, with Microsoft 365 as the initial workload. Intelligent ResOps uses the graph to unify data, context and recovery across SharePoint, OneDrive, Teams and Exchange. When an AI assistant or human makes a bad change, teams can see exactly what changed and whether the data is sensitive or regulated, then “restore only what’s needed instead of broad, disruptive restores.”

Finally, Veeam launched a Data and AI Trust Maturity Model, developed with McKinsey and informed by input from more than 300 chief information officers and chief information security officers, to provide enterprises with a structured way to benchmark and plan their path from AI experimentation to demonstrable, auditable trust. Organized around the four pillars of Understood, Secured, Resilient and Unleashed, the model includes 12 dimensions and 49 subdimensions across five maturity levels. It is delivered as a consultative assessment featuring scored profiles, peer benchmarking and a prioritized roadmap.

Eswaran and Jalil talk AI challenges

Eswaran consistently returned to the theme that “recovery is the ultimate currency” in a world where AI, identity, and data are “completely connected, always under attack.” At the same time, he acknowledged that traditional notions of recovery are no longer sufficient: “You cannot roll back the enterprise” every time an agent goes wrong; instead, “remediation needs to be precise. You must be able to undo just those five seconds of agent action, just that one element in a file which got changed.”

He also emphasized that Veeam’s rebranding as “the Data and AI Trust Company” is more than label-swapping. “For us, DataAI is not just a branding exercise; it is where data, access and controls, identity and AI come together in one connected platform, because in the agentic era, you cannot solve these as individual problems.”

Jalil, whose Securiti team now forms the core of Veeam’s data security and governance stack, framed the opportunity and responsibility for resilience teams in the AI era. “If you’re trying to bring AI into the enterprise, you’re not going to put your intellectual property and your information behind it without guardrails and without knowing that you can recover from anything,” he said. “That really is the opportunity for our community and for us to play a central role in enabling the safe transformation toward AI.”

Why this pivot matters

From an industry-structure standpoint, Veeam’s move expands the competitive field it operates in. The company is no longer just competing against legacy backup vendors; it’s now colliding with DSPM providers, identity-centric security platforms, privacy automation tools and a growing wave of AI governance startups, while also claiming a unique position as the only vendor that deeply understands both the live data plane and the backup plane.

For customers, the question is whether it’s better to assemble trust capabilities from multiple best-of-breed tools or to consolidate on a unified platform that unifies data, identities, AI agents, and resilience. Veeam’s bet is that in the agentic era, running security, governance, compliance, privacy, and recovery as separate disciplines with different vendors, budgets, and UIs “stops being an option,” because every gap in context becomes a gap in trust.

That’s the essence of Veeam’s big pivot: from promising to get you back up when “everything else breaks” to promising that your data, and the AI acting on it, will be understood, governed and recoverable by design. If the agentic era unfolds as Eswaran and Jalil describe it, the companies that can operationalize that promise at scale will define the next infrastructure category.

If Veeam had acquired a cyber company five years ago, the industry might have said, “Huh? Why?” and viewed Veeam as a company with a solution looking for a problem. However, operating AI at scale is fundamentally different from operating the information technology environment of a decade ago. Make no mistake, the agentic era is coming fast, and it’s going to create problems IT leaders can’t comprehend. In an analyst Q&A, Eswaran discussed Veeam’s competitive position: “Regarding that 80% of the Fortune 500 – our forward-looking thesis centers on the ability to leverage a ‘Contextual Intelligence Knowledge Graph.’ This provides essential context to your data and its surrounding ecosystem, bridging relationships across identity, AI, and agents. We believe this will be our primary differentiator.”

Final thoughts

For customers, the takeaway from Veeam’s big pivot is to treat data and AI trust as an architectural requirement, not an add-on feature. That means inventorying where AI agents already touch your critical data, consolidating visibility across production and backup, and insisting on controls that can both prevent bad actions and surgically unwind them when they occur.

Don’t wait for a regulatory mandate or a headline-making incident to force the issue; use this moment to pressure-test your identity, governance, and recovery assumptions against an “assume autonomy” world. The organizations that move now to unify security, compliance, privacy and resilience around a common data graph will be the ones that can adopt AI fastest — because they’ll be the only ones that can prove, to themselves and to others, that they can trust it.

Zeus Kerravala is a principal analyst at ZK Research, a division of Kerravala Consulting. He wrote this article for SiliconANGLE.

Photo: Zeus Kerravala