Cohesity has been granted U.S. Patent No. 12,619,501 by the U.S. Patent and Trademark Office (USPTO) for its generative AI retrieval-augmented generation (RAG) platform built on secondary data. Issued May 5, 2026, and titled "Data Retrieval Using Embeddings for Data in Backup Systems," the patent marks a significant development for enterprises seeking to leverage backup data for AI without moving it. The patent covers the foundational technology behind Cohesity Gaia, enabling organizations to apply GenAI directly to secondary data while maintaining security, governance, and access controls.
Cohesity's Industry-First Patent for Secondary Data RAG
The patent validates Cohesity's architectural approach to combining secondary data systems with a RAG semantic layer for GenAI applications. This proprietary method enables enterprises to apply AI to backup data without creating new data silos, weakening governance controls, or increasing the exposure of sensitive information. Cohesity is the first data protection vendor to patent this approach, which keeps data protected, governed, and in place during AI processing. The patent was invented by Gregory Statton, Sanjay Poonen, Mohit Aron, and Apurv Gupta. This team spans Cohesity's engineering and executive leadership, reflecting the strategic importance of the underlying innovation to the company's long-term platform vision.
Security-First Architecture for Secondary Data AI
Cohesity Gaia leverages this patented technology to make secondary data—including files, emails, databases, and virtual machines—securely searchable and usable as a governed knowledge source for GenAI applications. Instead of requiring enterprises to replicate sensitive data into separate AI infrastructure, the approach enables AI workloads to run against protected secondary data while preserving existing security, governance, compliance, and access controls. By making this data semantically searchable and usable by large language models (LLMs) without requiring users to move or copy it, the platform minimizes risk. As Cohesity's leadership notes, protected recovery data is a "goldmine" and an organization’s most complete repository of institutional knowledge, yet it remains highly underutilized. This security-first architecture reduces data sprawl, limits the attack surface, and extends trusted data into AI workflows.
Enterprise Benefits and Customer Validation
The patent addresses the underutilization of high-value secondary data for AI use cases. By making this data semantically searchable without moving or copying it, organizations can unlock insights from years of historical data while maintaining cyber resilience. Patrick Ringelberg, Domain Data Center Architect & AI at the Netherlands' Ministry of Infrastructure and Watermanagement (Rijkswaterstaat), noted that preserving security posture and ensuring sovereign, on-premise control as a Dutch government institution were critical objectives. Ringelberg emphasized that Cohesity's approach was the only one that made AI viable using existing backup data as the foundation, reinforcing its differentiated nature. The platform enables enterprise teams to accelerate decision-making and build AI-driven workflows on a secure foundation that inherits the controls already governing their recovery environment.
Key Takeaways
- Cohesity secured U.S. Patent No. 12,619,501 for its GenAI RAG platform built on secondary data, issued May 5, 2026.
- The patent enables enterprises to apply GenAI to secondary data without moving or copying it, preserving security and governance controls.
- Cohesity Gaia, leveraging this technology, is available today as part of the Cohesity Data Cloud platform.
TechInsyte's Take
This patent underscores Cohesity's technical lead in applying GenAI to secondary data while maintaining security posture. While the approach offers a clear alternative to data replication for AI workloads, its broader market adoption and scalability across diverse enterprise environments remain to be seen. Buyers should monitor how Cohesity integrates this technology with existing data protection workflows and whether it becomes a standard for secure GenAI on secondary data.
Source: Businesswire
