Cummins treated information governance and generative AI adoption as a single, connected initiative. As the company piloted Microsoft 365 Copilot and joined the Microsoft Copilot Early Adoption Program, it upgraded from Microsoft 365 E3 to E5 to gain access to Microsoft Purview. With Microsoft Purview, Cummins: - Automated data classification and sensitivity labeling using Microsoft Purview Information Protection. - Applied persistent sensitivity labels so that confidential or restricted content stays protected across apps and locations. - Used data classification features to help ensure that responses created by generative AI tools like Microsoft Copilot are tagged as confidential or restricted when appropriate, so only the right people can see them. - Began evaluating Microsoft Purview AI Hub to gain visibility into AI interactions and to use audit trail logging for AI usage across the environment. By automatically labeling and classifying more than one million files and millions of email messages, Cummins created a cleaner, more secure data estate. This managed, “clean” data environment is what enables the company to adopt generative AI more safely and to get more accurate, policy-compliant AI outputs.

Cummins operates in more than 180 countries, with around 76,500 employees and over 100 years of accumulated data. Much of this data is unstructured content created in Microsoft 365—Word documents, PowerPoint files, Teams chats, Outlook emails, videos, and GIFs. This created several challenges: 1. **Complex retention and legal requirements** - Some records must be retained for the life of a product, which can span decades. - Different jurisdictions impose different legal and regulatory requirements. - Retaining too much data increased storage costs and made eDiscovery more complex and expensive. 2. **Fragmented records management and limited collaboration** - Historically, employees manually applied static security labels and moved important files into a separate electronic records repository. - While this helped with administration, it made collaboration difficult because employees could not easily co-author or update official records. 3. **Exponential data growth and multiple IT environments** - Tens of millions of files across multiple infrastructures (due in part to acquisitions) made consistent governance difficult. - Too much unclassified data also meant more noise for generative AI tools to sift through. Microsoft Purview helped Cummins address these issues by: - **Automating classification and labeling** with Microsoft Purview Information Protection, so sensitivity labels and protections are applied consistently and persistently. - **Implementing Purview Data Lifecycle Management** to assign customized lifecycles to documents, reduce outdated files, and cut storage and maintenance costs. - **Using machine learning** in Purview to automate 10–15% of the global record schedule and to support auto-tagging for more than 50 label types on email, with millions of messages auto-tagged in 2024. - **Centralizing governance across Microsoft 365** (email, SharePoint, and other services) so that classification, retention, and protection policies are integrated rather than managed in separate systems. As a result, Cummins streamlined data security and compliance, reduced manual effort, and improved collaboration by allowing employees to co-author “living” documents in the tools they already use, while governance and compliance run largely in the background.

Cummins is using Microsoft Purview to reshape how it manages information, with several measurable and operational benefits: 1. **Cost savings and efficiency** - By using Purview Data Lifecycle Management to enforce retention policies and customized file lifecycles, Cummins has reduced outdated files and associated storage and maintenance costs. - Automation of records management—currently 10–15% of the global record schedule—is reducing manual administration and freeing employees from routine data governance tasks. 2. **Scale and coverage across the data estate** - More than one million files have been labeled using Microsoft Purview. - Over 50 label types are in place for auto-tagging email, with millions of email messages auto-tagged in 2024. - Purview identifies confidential information using hundreds of prebuilt rules tailored to regions around the world, helping Cummins decide what to retain or delete. 3. **Improved security, compliance, and user experience** - Sensitivity labels are persistent and can include protection settings such as encryption, which define what authorized users can do with content inside and outside the Cummins domain. - Automation reduces the risk of human error, while still allowing employees to manually adjust labels when needed. - Employees can now co-author and update sensitive files directly in Microsoft 365 apps, instead of working in separate repositories, improving collaboration without sacrificing control. 4. **A stronger foundation for AI and knowledge management** - Clean, well-classified data supports safer and more accurate use of generative AI tools like Microsoft 365 Copilot. - With Microsoft Purview AI Hub, Cummins plans to gain visibility into AI interactions and use audit trails to support secure AI adoption. - Over time, Cummins expects Purview to evolve from a governance and protection platform into a broader knowledge management solution, helping employees quickly find the most accurate, role-appropriate information. Overall, Cummins is using Microsoft Purview to reimagine information governance—reducing costs, improving compliance, and creating a more AI-ready, collaborative data environment.