Veeam has officially announced that Intelligent ResOps (iResOps) for Microsoft 365 is becoming a General Availability (GA) product. This is the first iResOps solution to reach GA status since the concept was introduced in May. The new feature, file-by-file undo, allows administrators to revert individual files that have been modified by artificial intelligence (AI) agents without having to restore the entire backup.
iResOps is an add-on to Veeam Data Cloud for Microsoft 365, so it connects to your existing service without data migration or major architectural changes. The solution automatically associates a tenant’s AI agents, the sensitive data they can access, and the files they modify with the state of their backups. This provides visibility that has been lacking until now, as organizations often don’t know what AI tools are running in their environment.
AI agent visibility and sensitive data mapping
Veeam iResOps extends Veeam Data Cloud with the DataAI Command Graph, a centralized data map that maps production data from SharePoint, OneDrive, Teams, and Exchange to the context of their backups. In this graph, the solution discovers Copilot and other AI agents, shows which sensitive data they can access, and warns of potential oversharing. When an AI agent incorrectly modifies or deletes files, administrators can review the changes at the file level and roll back each one, avoiding a large data restore operation.
Restore point and recovery map
„Veeam Intelligent ResOps provides customers with both a recovery point and a recovery map. A recovery point shows what can be recovered, and a recovery map shows what to recover first, what has changed, and what no longer needs to be recovered,“ said Anand Eswaran, CEO of Veeam. The map is powered by Securiti AI technology, which Veeam acquired and integrated into its DataAI Command Platform. In addition, iResOps classifies sensitive and regulated data across all four Microsoft 365 workspaces, shows who can access it and how, and identifies redundant, obsolete, or trivial (ROT) data that can be quarantined or archived.
Minimal recovery planning and automated insights
The solution helps confirm that business-critical data is backed up, uncovers security gaps before an incident occurs, and identifies which data is most critical for business continuity. The tool also answers simple questions in natural language about data, access, AI agents, and backup protection. Veeam Data Platform Preview lets you connect existing backup servers to the cloud-based DataAI Command Platform without data migration, while Command Graph connects protected infrastructure, workloads, and indexed files in a single console.
Additional iResOps options and future plans
In addition to Microsoft 365, Veeam has opened private previews of iResOps for Veeam Data Platform and Okta. iResOps for Okta is a managed services solution for identity misconfiguration and malicious changes. Because Okta does not have native point-in-time recovery or version history, iResOps creates immutable recovery points for all users, groups, applications, policies, and admin roles, allowing you to restore objects to a specific property level without having to restore the entire tenant.
Availability and licensing
Veeam Intelligent ResOps for Microsoft 365 is available today as an add-on to Veeam Data Cloud for Microsoft 365 in both Foundation and Advanced editions, as well as through Veeam Cloud & Service Provider partners. Subsequent versions, such as Veeam Data Platform v13.2, are scheduled to become GA later in Q4 2026.
Conclusion
Veeam iResOps gives organizations visibility, control, and flexibility when working with AI agents in Microsoft 365 environments. The ability to recover files one at a time, along with data mapping and ROT data management, helps reduce the risk of incorrect AI agent actions and optimize storage costs. This is an important step forward in data resiliency and governance, especially given the Veeam Data Trust & Resilience Report 2026, which highlights that the majority of organizations are already using or testing AI agents, but only a small percentage are AI-ready.
