Cummins recognized that generative AI is only as useful as the data it can safely access. With more than 100 years of data, operations in 180+ countries, and approximately 76,500 employees, the company needed to ensure that AI tools would work with clean, well-governed information.

Several factors drove this decision:

• Regulatory and retention requirements: Some records must be kept for the life of a product, which can span decades. Cummins needed proper record labeling and retention policies to stay compliant across multiple jurisdictions.
• Unstructured data growth: Most data was unstructured content in Microsoft 365 (Word, PowerPoint, Teams chats, Outlook emails, videos, GIFs). This made it hard to classify, secure, and manage at scale.
• AI accuracy and safety: The team understood that Microsoft 365 Copilot needed secure, clean data to generate accurate responses. Poorly governed data would increase risk and reduce the value of AI.
• Cost and complexity: Retaining tens of millions of outdated files increased storage costs and made eDiscovery and AI processing more complex and expensive.

By implementing Microsoft Purview first, Cummins could automate classification, labeling, and retention, so that generative AI would operate on trusted, well-protected data from day one.

Cummins adopted Microsoft Purview as the foundation for its information governance program and reimagined how records are managed across the business.

Key actions included:

• Upgrading to Microsoft 365 E5: This gave Cummins access to the full Microsoft Purview suite for security, governance, and compliance.
• Automating data classification and sensitivity labeling: Using Microsoft Purview Information Protection, Cummins automatically classifies and labels content so that sensitivity levels and protections (such as encryption and access restrictions) are consistently applied.
• Implementing Data Lifecycle Management: With Purview Data Lifecycle Management, the company assigns customized retention lifecycles to documents, helping reduce excess and outdated records and associated storage costs.
• Scaling labels and auto-tagging:
  • More than 50 label types are now used for auto-tagging email messages.
  • Millions of email messages were auto-tagged in 2024.
  • Over one million files have been labeled using Microsoft Purview.
• Using machine learning for records management: Cummins has automated 10–15% of its global record schedule with Purview’s machine learning capabilities and plans to extend this with generative AI.

This integrated approach lets Cummins manage data across the entire Microsoft 365 environment, reduce manual effort, and lower storage and maintenance costs while keeping governance and compliance controls in place.

Before Microsoft Purview, Cummins relied on static security labels and a centralized records repository. Important files were moved into that repository, which helped with control but made collaboration difficult because employees could no longer easily edit official records.

With Microsoft Purview, Cummins reshaped this model:

• Persistent sensitivity labels in everyday tools: Sensitivity labels now travel with the document, so employees can work in familiar apps like Word, PowerPoint, Outlook, Teams, and SharePoint while protections remain in place.
• Actionable classification: When a document is labeled as confidential, its access and handling rules are automatically enforced inside and outside the Cummins domain. Employees can apply labels manually, or rely on auto-classification policies running in the background.
• Integrated security and collaboration: Instead of locking content away in separate systems, employees co-author “living documents” in a single environment that combines collaboration, security, and automated compliance.
• Support for safe AI use: Cummins uses Purview’s data classification features so that responses generated by tools like Microsoft 365 Copilot can be tagged as confidential or restricted, ensuring that only the right people see sensitive AI-generated content.
• Visibility into AI interactions: The company is piloting Microsoft Purview AI Hub to gain visibility into Copilot and other AI interactions, and plans to test audit trail logging to track AI usage across the environment.

As a result, Cummins can collaborate more effectively across its global workforce while maintaining consistent security, compliance, and a data foundation that is ready for broader generative AI adoption.