Rubrik Agent Rewind allows businesses to reverse errors caused by autonomous AI

Rubrik has launched Agent Rewind, a new solution developed following its acquisition of Predibase. This innovative tool, powered by Predibase’s AI infrastructure, allows organisations to reverse mistakes made by agentic AI. By providing enhanced visibility into the actions of AI agents, Agent Rewind enables enterprises to trace, audit, and safely revert changes made to applications and data. Johnny Yu, Research Manager at IDC, emphasised the importance of addressing the potential for ‘non-human error’ in AI systems, urging organisations to consider solutions that can mitigate the risks associated with AI mistakes. Anneka Gupta, CPO at Rubrik, noted that as AI agents become more autonomous, the likelihood of unintended errors increases, which can lead to significant business disruptions.

Agent Rewind addresses these challenges by making previously opaque AI actions visible and auditable. It creates an audit trail and immutable snapshots that facilitate safe rollbacks, distinguishing itself from current observability tools that only report what happened without explaining how to reverse high-risk actions. Chad Pallett, CISO at BioIVT, highlighted the necessity for observability and secure rollback in AI usage, stating that Agent Rewind will provide not only data safety and model speed but also AI recoverability. The solution offers context-enriched visibility, mapping agent behaviour back to its root cause, and integrates seamlessly with various platforms and APIs, including Agentforce and Microsoft.