Microsoft Builds Synthetic Enterprise Environments to Test Copilot at Scale

Microsoft Research has launched the Industry Evaluation Data Generation Platform, a system that creates fully synthetic enterprise environments for testing its Copilot AI. The platform draws on publicly available data such as O*NET occupational data, online job postings, and SEC filings to generate realistic users, meetings, emails, documents, and organizational structures. Automated query pipelines produce reproducible evaluation datasets, while quality frameworks ensure the synthetic data meets rigorous standards. By using these synthetic scenarios, Microsoft can assess Copilot’s performance—such as summarizing meetings and extracting action items—without accessing any real customer data, thereby addressing privacy and safety concerns while providing a more realistic benchmark than generic tests.